Let's discuss why AMD is no longer an enthusiast platform

[Soarin]

You don't have to. The 3570K can be had for $229.99 on Newegg or around $199.99 at Microcenter if you have one near you. The applications where Vishera might be a better option are few and far between for the home user or gamer. The 3770K is $329.99 on Newegg or can be had for around $299.99 at Microcenter. Though it's $300 it won't be beaten by Vishera at hardly anything.

Ah well I did not realize that so my post be least invalid, or somewhat un-thought out or just from sleep deprivation from yesterday and other events..

 

[Dan_D]

But it isn't gnoreeverythingelsethatdoesntrevolvearoundgaming|OCP.

I mentioned the SB-E in the post, but it's also much more expensive. If you think about it, what normal usage would the chip be insufficient in other than slip behind in gaming?

I don't ignore everything but gaming. I know the site is focused largely in that direction but I do not totally ignore non-gaming applications. In threads I'm often the one that brings up usage aside from gaming. Again I never said it would be insufficient for normal usage. I'm just saying that there are times where performance does justify cost.

It isn't a matter of being totally insufficient. But again sometimes performance matters. When rendering scenes in 3D Studio Max, it matters. A few seconds can turn into hours as that time adds up quickly. If you have 1,000 frames to render one night and one chip does it in 45 seconds per frame vs. 60 seconds, that can be a big deal. It can be the difference between setting a render to run over night and it being complete by the time you arrive in the morning vs. waiting an hour or two for it to finish.

Now examples of time crunches that can be processor dependant are few and far between. It's just an example. On the gaming side you have to get into high end multimonitor and multiGPU configurations for it to matter. But again a few FPS at those resolutions is the difference between playable vs. unplayable.

If you're going to benchmark then benchmark with actual applications that people use. That's exactly what Tom's did. Using synthetic benchmarks alone is akin to measuring penis sizes according to how quickly it runs cinebench but avoiding a ruler.

And those benchmarks from Tom are largely being ignored here. Again they show Vishera losing to Ivy Bridge by a small margin. But look at the context. These are the heavily threaded applications which people often speak of to justify their love of AMD. And Vishera is getting beaten by a processor with half as many cores with a 300 or 400MHz clock speed deficit. Cost wise that's not too bad for Vishera as it's getting the job done but as an enthusiast these stock clocks aren't what matters. Ivy Bridge should clock better than Vishera and it already has an IPC advantage.

That's just me I guess. But as an enthusiast a CPU getting beaten by one with lower clocks, half as many cores with some fake virtual processors on it is pretty lame.

Because it's about both price and performance. I'm not buying a 3960X because it's the best chip on the planet because it costs a bagillion dollars. Likewise, there's clearly an advantage in opting to go with a 8350 over a 3770K if you can spend the ~$120 on a GPU or RAM or what have you. Ultimately you're going to get a better rig if your cash is evenly spread out amongst the parts.

I disagree. If you spend $100 to $150 more you'll get a much faster rig overall. If that $150 comes out of your video card budget alone, and gaming is your priority, sure. I get it. And these aren't really expensive Intel CPUs either as I illustrated above. Yes they cost a little more than the AMD FX-8350 does but when building a whole computer $100 or $150 isn't that big a stretch. I know for some people it could be but I'd just save my pennies another week and get what I really wanted.

 

[CleanSlate]

I disagree. If you spend $100 to $150 more you'll get a much faster rig overall. If that $150 comes out of your video card budget alone, and gaming is your priority, sure. I get it. And these aren't really expensive Intel CPUs either as I illustrated above. Yes they cost a little more than the AMD FX-8350 does but when building a whole computer $100 or $150 isn't that big a stretch. I know for some people it could be but I'd just save my pennies another week and get what I really wanted.

One thing I've seen in the past is that Intel tends to have rather expensive motherboards in comparison to AMD though, so you can't forget the early adopters who have to deal with $300+ motherboards for the same features as an AMD that's around $150-$200.

 

[pelo]

And those benchmarks from Tom are largely being ignored here. Again they show Vishera losing to Ivy Bridge by a small margin. But look at the context. These are the heavily threaded applications which people often speak of to justify their love of AMD. And Vishera is getting beaten by a processor with half as many cores with a 300 or 400MHz clock speed deficit. Cost wise that's not too bad for Vishera as it's getting the job done but as an enthusiast these stock clocks aren't what matters. Ivy Bridge should clock better than Vishera and it already has an IPC advantage.

They're not exactly pulling out weird applications which nobody uses. I'd agree with you if it was some weird ass application I've never even heard of, but CS6? Sony Vegas? Blender? 3dx Max? Of course they all make AMD's architecture better. They're productivity applications which are always going to be updated with newer ISAs and parallelized. It's the one place where moar coars = better.

That's just me I guess. But as an enthusiast a CPU getting beaten by one with lower clocks, half as many cores with some fake virtual processors on it is pretty lame.

I don't care so long as it gets the job done. Do you also consider the Fermi architecture better than Kepler because of the core count and clock speeds? Is one of those not as enthusiast-y?

I disagree. If you spend $100 to $150 more you'll get a much faster rig overall. If that $150 comes out of your video card budget alone, and gaming is your priority, sure. I get it. And these aren't really expensive Intel CPUs either as I illustrated above. Yes they cost a little more than the AMD FX-8350 does but when building a whole computer $100 or $150 isn't that big a stretch. I know for some people it could be but I'd just save my pennies another week and get what I really wanted.

This can go into a monitor, bumping up to IPS. It can go into an SSD. A GPU. Whatever. Anyway, that's not the point...

People build rigs on a budget. If it's going to be productivity focused, you're probably better off AMD. This shouldn't be surprising as we saw similarities with the Thubans as well. If you're going to be gaming, go for a 2500K. If you can find scrap together more pennies then buy the 3770K. If you can scrap together even more pennies go with 2011. The point here is that it still fits a niche; though I'd argue their 2 and 3 module parts + Trinity fit a much larger niche than do the 4 module chips.

Even if AMD made an amazing desktop CPU they'd still be in deep shit. Same goes for Intel, too.

 

[Dan_D]

One thing I've seen in the past is that Intel tends to have rather expensive motherboards in comparison to AMD though, so you can't forget the early adopters who have to deal with $300+ motherboards for the same features as an AMD that's around $150-$200.

People always say this but there are plenty of $200 LGA1155 boards.

They're not exactly pulling out weird applications which nobody uses. I'd agree with you if it was some weird ass application I've never even heard of, but CS6? Sony Vegas? Blender? 3dx Max? Of course they all make AMD's architecture better. They're productivity applications which are always going to be updated with newer ISAs and parallelized. It's the one place where moar coars = better.

And yet the benchmarks show that isn't enough. Vishera is still second place most of the time to a processor with half as many cores. Yes more cores are better, but AMD is still pulled down by their relatively poor IPC performance in those same multithreaded applications which put AMD's Vishera in a better light.

I don't care so long as it gets the job done. Do you also consider the Fermi architecture better than Kepler because of the core count and clock speeds? Is one of those not as enthusiast-y?

I consider the one that's faster to be the better option all things being equal. Fermi and Kepler aren't directly comparable in clocks because of architectural differences. In the case of Vishera vs. Ivy Bridge that's not the case. They are more directly comparable. Their clocks are closer and we can match them up in the same applications at the same clocks and see how they do. And again Ivy Bridge can clock higher and is faster per clock cycle anyway. I don't know why some people are having a hard time understanding that.

This can go into a monitor, bumping up to IPS. It can go into an SSD. A GPU. Whatever. Anyway, that's not the point.

That's up to the person doing the building / ordering.

People build rigs on a budget. If it's going to be productivity focused, you're probably better off AMD. This shouldn't be surprising as we saw similarities with the Thubans as well. If you're going to be gaming, go for a 2500K. If you can find scrap together more pennies then buy the 3770K. If you can scrap together even more pennies go with 2011. The point here is that it still fits a niche; though I'd argue their 2 and 3 module parts + Trinity fit a much larger niche than do the 4 module chips.

Even if AMD made an amazing desktop CPU they'd still be in deep shit. Same goes for Intel, too.

From a bang for your buck perspective, perhaps. From a pure performance standpoint, it would seem that's not the case. Again in the benchmarks you posted Vishera is getting beaten by a 3770K in all those multithreaded applications. For some reason you seem to ignore that fact. And it is doing so with a clock speed deficit. It will clock better and perform better than Vishera. As you overclock them that gap will probably widen. If you are really serious about it an LGA2011 processor based system would absolutely destroy Vishera in those same applications as it has more cores, higher clocks and better IPC. Also some of those applications do benefit from added memory bandwidth. Something Intel has in spades over AMD right now.

 

[pelo]

From a bang for your buck perspective, perhaps. From a pure performance standpoint, it would seem that's not the case. Again in the benchmarks you posted Vishera is getting beaten by a 3770K in all those multithreaded applications. For some reason you seem to ignore that fact. And it is doing so with a clock speed deficit. It will clock better and perform better than Vishera. As you overclock them that gap will probably widen. If you are really serious about it an LGA2011 processor based system would absolutely destroy Vishera in those same applications as it has more cores, higher clocks and better IPC. Also some of those applications do benefit from added memory bandwidth. Something Intel has in spades over AMD right now.

If there's a $120 delta, that's a pretty important point, no? I'd rather have a beefy GPU than a beefy CPU. In fact, that goes for most of the people on this site. $120 SSD or an Intel quad+HT? Why not an Intel quad + SSD?

I mean, take a look at the [H] review, ffs.

Once again with this synthetic calculation benchmark, that is singled threaded as we have it set up, we see Vishera come in behind Zambezi and greatly trail the Intel architectures.

Great... and what does that mean? Why is that important? What is that synthetic benchmark testing with respect to the architecture? Nothing? Something? Everything? If you go by that review, you'd have no fucking idea.

So let's give a reason here, shall we?

That Pi benchmark is x87 hardware accelerated on Intel chips whereas AMD lacks x87 hardware. They removed it post-K8. This dates back to the 8087; we're talking ages. In modern software, or pretty much any software outside of that particular benchmark, you're not going to see this. Instead you'll be greeted with AVX, FMA and SSEx.x, with the latter still the most popular. Yet there it is in [H]'s review to showcase the lower IPC. Of course, it doesn't bother telling you why it's completely insignificant, but it's still makes a nice useless graph. In short, Intel's first-gen Core i3's would still stomp on AMD's Vishera chips clocked at 5ghz. Are they better processors?

Here's a question: Do you even want x87 hardware acceleration? Can you point to somewhere it makes sense to have it?

The answer to that is almost certainly no, and I can't to the latter. So why the benchmark?

Yet you're telling me that Tom's has an AMD favored heavily-threaded benchmark suite? Dan, at least their benchmarks were useful.

And those benchmarks from Tom are largely being ignored here. Again they show Vishera losing to Ivy Bridge by a small margin. But look at the context. These are the heavily threaded applications which people often speak of to justify their love of AMD.

No, they used those benchmarks because those are actual applications people use. Claiming favoritism is ridiculous if you actually bothered to read their conclusion. Because they concluded the same thing I'm pointing out here: It makes sense as a budget productivity chip. Is that really hard to comprehend? Did we not see the exact same with the Thuban?

The 3770K is without a doubt the better processor, but it's also without a doubt $120 more expensive. It's without a doubt more power efficient, but if you're looking for lower power consumption you should be paying attention to motherboards and idle power than full load power consumption (unless you're folding). The 3570K is bar none the best gaming processor and the best overall processor, but it slips behind the comparatively priced 8350 in productivity software.

It clearly has a niche. It's an incredibly small niche, but it's still there. The 6300 and 4300 are much better buys, though, knocking off the i3's off pretty much everyone's list of recommendations.

This puts AMD in the exact same spot they were when they had the Thuban and Deneb lines. Great bang for the buck but not enough for the high end. AMD has been in that position since Conroe

 

[CleanSlate]

People always say this but there are plenty of $200 LGA1155 boards.

Now there are plenty of good boards out there but my point was that especially during the time when the 1100t was in full swing, the list of decent Intel boards could fit on less than one hand and their prices started at $300 or more. I'm talking about dual x16 pcie slots with one or two usb 3.0 slots, enough usb slots in general, and enough sata slots for any one that actually wants to use their system like an enthusiast (or just any one smart).

This went on for about a year, I believe. I'm not entirely sure why but that's what I saw and I had to deal with that as a consumer and frankly my opinion of Intel as a company has formed around that in some portion or another.

So, in the future, I wouldn't doubt this would be the case again due to licensing or whatever the case may be.

 

[Blkout]

The thing about high end bleeding edge performance is that the R&D that it takes to get there allows a company to learn from it and use that technology to trickle down into lower end products from the high end parts. There's a reason why even Intel's low end products are still very good performers, its because they use the same or smilar technology as their high end parts just cut down.

 

[CleanSlate]

The thing about high end bleeding edge performance is that the R&D that it takes to get there allows a company to learn from it and use that technology to trickle down into lower end products from the high end parts. There's a reason why even Intel's low end products are still very good performers, its because they use the same or smilar technology as their high end parts just cut down.

YES, someone gets exactly what I'm trying to say.

Bleeding edge tech doesn't just find its way into the $1000 cpus. If a company can't be great at least at one thing, then why buy any thing they make at all? What's the incentive?

 

[Blkout]

And this is exactly the reason why AMD can't get back to the top again, as long as their top tier products are only good enough to be released as "bang for the buck" products, the profit in those products is slim and AMD doesn't sell enough of these "bang for the buck" CPU's to be profitable in that small market which is why the continue to bleed money. Its not like AMD has chosen to target this market, they have no choice but to live in that market because they don't have the resources anymore to get back to the top and start being competitive. So they turn out their best effort which turns into a budget "bang for the buck" CPU which no OEM really wants. This is exactly why AMD is going to fall, its not if, but when. Its a slow death and upper managment probably knows it too but are riding that train until it finally derails. No one wants to admit it, but its already GAME, SET, MATCH, Intel.

 

[Dan_D]

If there's a $120 delta, that's a pretty important point, no? I'd rather have a beefy GPU than a beefy CPU. In fact, that goes for most of the people on this site. $120 SSD or an Intel quad+HT? Why not an Intel quad + SSD?

I mean, take a look at the [H] review, ffs.

Great... and what does that mean? Why is that important? What is that synthetic benchmark testing with respect to the architecture? Nothing? Something? Everything? If you go by that review, you'd have no fucking idea.

So let's give a reason here, shall we?

That Pi benchmark is x87 hardware accelerated on Intel chips whereas AMD lacks x87 hardware. They removed it post-K8. This dates back to the 8087; we're talking ages. In modern software, or pretty much any software outside of that particular benchmark, you're not going to see this. Instead you'll be greeted with AVX, FMA and SSEx.x, with the latter still the most popular. Yet there it is in [H]'s review to showcase the lower IPC. Of course, it doesn't bother telling you why it's completely insignificant, but it's still makes a nice useless graph. In short, Intel's first-gen Core i3's would still stomp on AMD's Vishera chips clocked at 5ghz. Are they better processors?

Here's a question: Do you even want x87 hardware acceleration? Can you point to somewhere it makes sense to have it?

The answer to that is almost certainly no, and I can't to the latter. So why the benchmark?

Yet you're telling me that Tom's has an AMD favored heavily-threaded benchmark suite? Dan, at least their benchmarks were useful.



No, they used those benchmarks because those are actual applications people use. Claiming favoritism is ridiculous if you actually bothered to read their conclusion. Because they concluded the same thing I'm pointing out here: It makes sense as a budget productivity chip. Is that really hard to comprehend? Did we not see the exact same with the Thuban?

The 3770K is without a doubt the better processor, but it's also without a doubt $120 more expensive. It's without a doubt more power efficient, but if you're looking for lower power consumption you should be paying attention to motherboards and idle power than full load power consumption (unless you're folding). The 3570K is bar none the best gaming processor and the best overall processor, but it slips behind the comparatively priced 8350 in productivity software.

It clearly has a niche. It's an incredibly small niche, but it's still there. The 6300 and 4300 are much better buys, though, knocking off the i3's off pretty much everyone's list of recommendations.

This puts AMD in the exact same spot they were when they had the Thuban and Deneb lines. Great bang for the buck but not enough for the high end. AMD has been in that position since Conroe

Why the hell are you arguing with me? What are you arguing about? I have been in agreement with the above and have made that point. I've said AMD makes sense in CERTAIN situations and not in others and that Intel has the overall better processor. Even benchmarks which show Vishera in a better light still show it losing. What the hell is so hard to understand about what I've said?

Now there are plenty of good boards out there but my point was that especially during the time when the 1100t was in full swing, the list of decent Intel boards could fit on less than one hand and their prices started at $300 or more. I'm talking about dual x16 pcie slots with one or two usb 3.0 slots, enough usb slots in general, and enough sata slots for any one that actually wants to use their system like an enthusiast (or just any one smart).

This went on for about a year, I believe. I'm not entirely sure why but that's what I saw and I had to deal with that as a consumer and frankly my opinion of Intel as a company has formed around that in some portion or another.

So, in the future, I wouldn't doubt this would be the case again due to licensing or whatever the case may be.

No it didn't. I've been reviewing boards for a long time and within the first month or so there are lower priced boards available. Sometimes even at launch week. Though availability is hit and miss they did and do exist.

And this is exactly the reason why AMD can't get back to the top again, as long as their top tier products are only good enough to be released as "bang for the buck" products, the profit in those products is slim and AMD doesn't sell enough of these "bang for the buck" CPU's to be profitable in that small market which is why the continue to bleed money. Its not like AMD has chosen to target this market, they have no choice but to live in that market because they don't have the resources anymore to get back to the top and start being competitive. So they turn out their best effort which turns into a budget "bang for the buck" CPU which no OEM really wants. This is exactly why AMD is going to fall, its not if, but when. Its a slow death and upper managment probably knows it too but are riding that train until it finally derails. No one wants to admit it, but its already GAME, SET, MATCH, Intel.

And this is the painful point I've made time and time again.

 

[CleanSlate]

No it didn't. I've been reviewing boards for a long time and within the first month or so there are lower priced boards available. Sometimes even at launch week. Though availability is hit and miss they did and do exist.

Possibly, I was going by only one retail location of a Fry's at the time. The build I did at the time was an impulse buy but I was at that location a few times over that year and it didn't get any better.

 

[cageymaru]

Well my 8120 has held up very well in things that I do. My nephew wants to upgrade whenever the next round of Intel chips come out, but only because he reads [H] and wants to be faster in benchmarks. I might buy a Vishera just to have something to mess with but I doubt it as I already have a 8120. I know Intel is faster, but there really aren't applications that require more performance. I wish there were more applications that were multithreaded or made my PC cry, but I haven't encountered them.

I'll consider Intel whenever the successor to the 3000 series comes out. My next upgrade for my PC is to get better airflow through my case so I can maintain higher OC's on my hardware in general.

 

[Comixbooks]

Once modern PC games stop running on AMD chips I'll stop buying them.
I mean if a game is CPU Dependant and it affect the frame rate till unplayable.

 

[Dan_D]

Possibly, I was going by only one retail location of a Fry's at the time. The build I did at the time was an impulse buy but I was at that location a few times over that year and it didn't get any better.

Well retailers are hard to go by. Sometimes they have stock early and sometimes they won't catch up for a week or more. Newegg and Amazon are better indicators of availability than local retailers tend to be. Your right in that there aren't a massive amount of choices for the first week or so but there are usually one or two models. ASUS for example tends to launch several boards in a given series at once. There is always a stripped model, a couple of in betweens, a higher end board and almost one ROG board at the launch of a chipset it seems. MATX and other form factors tend to filter in over the course of a couple weeks. Sometimes one MATX is available on launch.

Once modern PC games stop running on AMD chips I'll stop buying them.
I mean if a game is CPU Dependant and it affect the frame rate till unplayable.

And for most people, you can't tell the differnce. I've enjoyed playing games on AMD processor based machines when I've used computers that weren't mine. I was hugely impressed with the first Llano based notebooks when they came out. You do get a few FPS more on the Intel side but whether or not that's worth it either depends on your priorities and how much cash you either have to spend or are willing to spend combined with your circumstances and usage scenario. In my case, at the resolutions I'm running a few FPS makes all the difference. It can be the difference between running a game on the highest settings or it being unplayable on the highest settings. To me that's worth some extra coin. But this really only happens when going with multiple monitors and graphics cards as I've said before.

But that is far from the norm.

 

[HvyMtl]

Ok, AMD is not focusing on the top 1-5% over achievers known as the enthusiast market. Big deal?

No. AMD should be focused on the profitable market. The enthusiast market = bragging rights, and marketing. It does not equal profit.
What then would be a profitable market? Well, IF you have to limit it to the desktop, the inexpensive "Average Joe" type computer bought by the typical consumer, and corporate settings.

The "Average Joe" computer needs to surf the web, do some light weight office applications (word processing mostly), and light gaming. AMD's APU concept fits this market nicely, as it is cheaper to buy the APU instead of a separate CPU and video card. This makes it cheaper to build by OEM.

Then, lets look at the Business side. Most businesses need office computers capable of office applications. Basically, word processing, and spreadsheets. Most desktops in the corporate sector are far from speed demons, and are designed to last a long time (5 years, give or take) as least expensive as possible. Graphics for gaming? Not needed. Again, the APU can still be used, as it is fast enough, and a cheap build for OEM. Again, no video card necessary.

So, if AMD can price a processor at low cost, and be competitive to Intel products on the low end, and show the cost savings to the OEM by going with the APU concept, they could profit well.

Besides, more and more "Average Joe" are buying low cost (under $400) laptops, instead of desktops now. This still points towards the APU, as it performs well enough, and it is less expensive than buying a graphics card. True, Intel has also seen the light, with their graphics incorporated on the chipset, not on the cpu, in the laptop world. This makes the laptop field more competitive.

In addition, both Intel and AMD need to address the tablet and cellphone processor niches. These too are profitable areas.

Win 8 focused on making everything act like a tablet, smacks of Microsoft wanting to get into this lucrative area without Intel. (Look at their own non-Intel processor tablets...) If AMD is smart, they could possibly work this to advantage...

 

[AeonF1]

Ok, AMD is not focusing on the top 1-5% over achievers known as the enthusiast market. Big deal?

No. AMD should be focused on the profitable market. The enthusiast market = bragging rights, and marketing. It does not equal profit.
What then would be a profitable market? Well, IF you have to limit it to the desktop, the inexpensive "Average Joe" type computer bought by the typical consumer, and corporate settings.

The "Average Joe" computer needs to surf the web, do some light weight office applications (word processing mostly), and light gaming. AMD's APU concept fits this market nicely, as it is cheaper to buy the APU instead of a separate CPU and video card. This makes it cheaper to build by OEM.

Then, lets look at the Business side. Most businesses need office computers capable of office applications. Basically, word processing, and spreadsheets. Most desktops in the corporate sector are far from speed demons, and are designed to last a long time (5 years, give or take) as least expensive as possible. Graphics for gaming? Not needed. Again, the APU can still be used, as it is fast enough, and a cheap build for OEM. Again, no video card necessary.

So, if AMD can price a processor at low cost, and be competitive to Intel products on the low end, and show the cost savings to the OEM by going with the APU concept, they could profit well.

Besides, more and more "Average Joe" are buying low cost (under $400) laptops, instead of desktops now. This still points towards the APU, as it performs well enough, and it is less expensive than buying a graphics card. True, Intel has also seen the light, with their graphics incorporated on the chipset, not on the cpu, in the laptop world. This makes the laptop field more competitive.

In addition, both Intel and AMD need to address the tablet and cellphone processor niches. These too are profitable areas.

Win 8 focused on making everything act like a tablet, smacks of Microsoft wanting to get into this lucrative area without Intel. (Look at their own non-Intel processor tablets...) If AMD is smart, they could possibly work this to advantage...

The flaw with focusing on low end is that the profit margin is minuscule compared to high end, especially if the cost of manufacturing a low end chip that isn't that much cheaper than making a high end chip. Sometimes they even cost the same to manufacture and the expensive chips are just better binned or the low end chip has a feature deactivated. Therefore, a higher priced processor gets more profit per chip than a low priced one.

The only way to succeed in the low end market is to sell a SHIT ton of chips which they haven't been doing. Compounded by the fact that AMD has to sell several chips just to match the profit margin of one Intel chip of the same price. With AMD being forced to sell at dirt cheap prices due to their performance compared to Intel, not to mention a 22nm chip is cheaper to make than a 32nm one due to yields, you have to wonder if AMD will live very long. If AMD can price their chips higher and still sell, they would.

 

[Dan_D]

The flaw with focusing on low end is that the profit margin is minuscule compared to high end, especially if the cost of manufacturing a low end chip that isn't that much cheaper than making a high end chip. Sometimes they even cost the same to manufacture and the expensive chips are just better binned or the low end chip has a feature deactivated. Therefore, a higher priced processor gets more profit per chip than a low priced one.

The only way to succeed in the low end market is to sell a SHIT ton of chips which they haven't been doing. Compounded by the fact that AMD has to sell several chips just to match the profit margin of one Intel chip of the same price. With AMD being forced to sell at dirt cheap prices due to their performance compared to Intel, not to mention a 22nm chip is cheaper to make than a 32nm one due to yields, you have to wonder if AMD will live very long. If AMD can price their chips higher and still sell, they would.

They also forget that the technology found in the higher end processors tends to trickle down to the rest of the product line. Minor improvements to your top end product makes the rest of the line better. When your best is basically the other guy's mid-range offering or worse, you are in trouble.

 

[Tsumi]

Well, AMD's approach was to make highly parallel chips great for server usage, which is where much of the money lies. Their problem is that for the most part power consumption was way out of control and clock speeds were too low. And of course, a highly parallel approach doesn't work on the desktop either.

 

[thomasward00]

For most people and even for most gamers the AMD FX CPU's are more than enough, especially at their pricepoints.

I just ordered my new build with an FX 6300, this will prob be my last pure CPU, by next gen both Intel and AMD will have gotten serious about APU's.

 

[ZodaEX]

When your best is basically the other guy's mid-range offering or worse, you are in trouble.

Not if your product is less than half the price.

 

[SLee]

Not if your product is less than half the price.

That's a negative for AMD, not a positive.

 

[Wascrash]

If anything AMD is an enthusiast platform. I can buy whatever I want. When BD came out I went AMD and now PD. They are alot of fun to overclock and really my PD can play anything out pretty dam fast and encode just fine even at stock settings. Thing is most here will want to overclock to make it faster than a stock i5/i7. Sure the overclocked Intel will still be faster but you wont notice unless your looking at benchmarks.

 

[Dan_D]

Not if your product is less than half the price.

This also means a lot less profit margin for AMD compared to Intel. Look at each company from a financial perspectice. AMD hardly has a winning strategy at the moment.

That's a negative for AMD, not a positive.

Agreed.

If anything AMD is an enthusiast platform. I can buy whatever I want. When BD came out I went AMD and now PD. They are alot of fun to overclock and really my PD can play anything out pretty dam fast and encode just fine even at stock settings. Thing is most here will want to overclock to make it faster than a stock i5/i7. Sure the overclocked Intel will still be faster but you wont notice unless your looking at benchmarks.

And we've covered this a dozen times. Or I have. The old socket is probably holding them back in a lot of ways. And you are wrong in that there are some instances where you would notice the difference outside of a benchmark. You might not go "goddamn cheap bastard must have an AMD processors" but rather "This thing could use a faster CPU."

 

[rumbl3]

I will say amd's current cpu offerings is nothing I'm willing to go jump onboard for. Not that they are not decent (amd microcenter deals free mobo is always a seller for me). Just in my case having my P2 Quad core I really have no reason to upgrade right now it plays everything great still and it just does not look like it would be a good upgrade unless I went with the newest hottest amd cpu $200.

I'm a amd fanboi tho I'll admit it. As long as there cpus stay nice and cheap with free mobos and keep at least medium performance I'll stick with them. Because I'm a cheap pc gamer usually buying mostly used parts and just price vs performance with a microcenter deal I walk away with a much better deal that way.

 

[Wascrash]

No Dont think I am wrong .Thats imo though. I do have an i5 here that I built for my son. Its not all that. BTW what instances are these?

 

[Wascrash]

No Dont think I am wrong .Thats imo though. I do have an i5 here that I built for my son. Its not all that. BTW what instances are these?
edit ..and what does this have to do with why AMD is not an enthusiast platform? They are fun to tinker with and can be made pretty dam fast when overclocked ..actually faster than the equivalent intel chip.

 

[Dan_D]

No Dont think I am wrong .Thats imo though. I do have an i5 here that I built for my son. Its not all that. BTW what instances are these?
edit ..and what does this have to do with why AMD is not an enthusiast platform? They are fun to tinker with and can be made pretty dam fast when overclocked ..actually faster than the equivalent intel chip.

The instances where AMD users might feel the pain is in high end, multiGPU configurations with multiple high resolution monitors. At 7680x1600 with 3 and 4-Way SLI AMD based machines will not produce the same FPS as Intel machines do. And the difference at that resolution can be the difference between what is playable and what is not. There are cases where multithreaded applications could benefit from reduced times. Render times which take all night could be painful if they are still running in the morning when you need the machine. And Intel is better there but admittedly at a much greater price. You need a 3770K there or step up to LGA2011.

And I can tinker all I want with just about anything. That by definition does not make it an enthusiast platform. A school bus isn't a really good platform for building a racecar off of. Yet I could do it. I can work on the engine. I could tinker with it all I wanted to. That doesn't make it a good idea or anything other than a money pit.

And I don't know in what fantasyland you think overclocked AMD's are faster than equivalent Intel's. Maybe on the super low end but not on AMD's upper "echelon CPUs". The FX-8350 is $220 and I've seen Core i7 3770K's go for $269.99 on sale and they will smack down anything AMD has at anything. And they'll clock just as high and have an IPC advantage over everything AMD's got.

 

[CleanSlate]

And I don't know in what fantasyland you think overclocked AMD's are faster than equivalent Intel's. Maybe on the super low end but not on AMD's upper "echelon CPUs". The FX-8350 is $220 and I've seen Core i7 3770K's go for $269.99 on sale and they will smack down anything AMD has at anything. And they'll clock just as high and have an IPC advantage over everything AMD's got.

Yea, this is the main problem.

 

[-Dragon-]

Seems to me a lot of the "AMD overclocks better/is faster when overclocked" is due in part to a resurgence in the whole "MHz myth" which, ironically enough, was (correctly) enspoused early on by the AMD crowd. They think an AMD at 5.0GHz is clocked over 10% faster than an intel at 4.5GHz there's no way that the IPC for the AMD could be that much worse than intels, never mind that a 1.8GHz Athlon XP stomped all over a 3.0GHz P4 with an over 50% clock advantage (or whatever it was back in the day, that seems right off the top of my head and I don't care enough to fact check, everybody here knows what I mean though). Basically the fantasy world they live on is one stuck in the 90's where MHz actually matter and so does what chip can reach the fastest clock speed on exotic cooling and all but one core shut down even though that has no real impact on day to day computing.

 

[Onemoar]

I do wish AMD would relax on the whole backwards compatibility bit
I would love to see AMD move to a LGA for there desktop chips Or maby they could invent something new
at the end of the day the whole argument comes down to price<>performance
is intels 10-15% advantage in performance worth the 40% difference in price ? illl go with no on this one
good enough is good enough unless you keep your system pegged at 100% load for hours on end then nobody will ever be the wiser
will a fx8150 @ ten cajillion giga hurtz ever be faster then intels chips that run at three point seven five mega hurtz ? NOPE
will the extra gigahurtz on amds side give you any performance boost .. maby
and people that have multi-gpu setups with 3 monitors don't buy amd they never have and never will so that point is moot
once you spend over a grand price pretty much becomes no object because you Obviously have money to blow

 

[Dangman]

at the end of the day the whole argument comes down to price<>performance
is intels 10-15% advantage in performance worth the 40% difference in price ? illl go with no on this one

Your argument happens to be poorly backed up. Which CPUs are we talking about here? 15% advantage in what applications? etc. In other words, expand on what you've said as what you just said doesn't make an ounce of sense without context.

 

[Pieter3dnow]

If you are a Linux user it is easy the CPU performs great, in Windows it does also when encoding. There still some drawbacks vs Intel but that is something most people don't mind.

The problem is that Windows and most of gaming is rather poorly written code. And I would say none of the benchmarks are optimized for AMD as they are for Intel unless you compile them yourselves (which people did with the x264).

But if the [H] definition of enthusiast is triple crossfire/SLI setups then yeah AMD falls short.. Also the chip is made for the server market more then anything else. We still don't know what Keller is doing at AMD if that will have a serious impact on Steamroller.

 

[-Dragon-]

Oh look, the "poorly written code" argument. Just because some program isn't threaded doesn't mean it's poorly written, many, many, MANY problems computers have to solve every day are single threaded because it's impossible to split them because further processing depends on previous processing to be finished before it starts. In (a + b) / c, you can't divide by c until a and b are finished adding, just like in a(b()) you can't start processing function a until b is finished. Servers don't care as much because they're doing lots of things at once so if you're running 20 threads on 8 cores, the 8 cores will stay nice and busy, but if it's a single user using 8 cores and the program has to calculate a single threaded solution before it can branch off, well, you find IPC matters.

 

[/dev/null]

The instances where AMD users might feel the pain is in high end, multiGPU configurations with multiple high resolution monitors. At 7680x1600 with 3 and 4-Way SLI AMD based machines will not produce the same FPS as Intel machines do. And the difference at that resolution can be the difference between what is playable and what is not. There are cases where multithreaded applications could benefit from reduced times. Render times which take all night could be painful if they are still running in the morning when you need the machine. And Intel is better there but admittedly at a much greater price. You need a 3770K there or step up to LGA2011.

And I can tinker all I want with just about anything. That by definition does not make it an enthusiast platform. A school bus isn't a really good platform for building a racecar off of. Yet I could do it. I can work on the engine. I could tinker with it all I wanted to. That doesn't make it a good idea or anything other than a money pit.

And I don't know in what fantasyland you think overclocked AMD's are faster than equivalent Intel's. Maybe on the super low end but not on AMD's upper "echelon CPUs". The FX-8350 is $220 and I've seen Core i7 3770K's go for $269.99 on sale and they will smack down anything AMD has at anything. And they'll clock just as high and have an IPC advantage over everything AMD's got.

I think the people who use 4 way sli/crossfiire with decent gpus couldn't care less about paying $150 or $300 for sometimes even $600 for a CPU.

People who are serious about rendering use render farms....

Btw, I have only bought a > $150 cpu once in the last 5 years and that was my 1090T. Still runs EVERYTHING I need and was ahead of it's time with iommu support.

I consider myself an enthusiast.

With that being said, a sub $150 intel CPU that has iommu support and much less power usage than my 1090T is probably my next purchase sometime in the next 3 years.

My current "work-from-home" desktop is a Pentium E5800 and the biggest upgrades I gave it wer

1) ssd's (it has 5 vertex1/agility1 ssds in an lvm raid0 volume+trim/ext4) and
2) an eyefinity 6 card so I can run 6 monitors (6870 2GB I got "returned" from CDW for $150).

 

[reb00tin]

WTF is an &#8220;enthusiast&#8221;? Introverted game-playing geeks that race their put0rs until they overheat and burn. Then they put plumbing in their put0r, so the mobo now burns at 6 GHhz instead of 5 GHz. And you think AMD should build CPUs for this trivial market share?

 

[SLee]

WTF is an “enthusiast”? Introverted game-playing geeks that race their put0rs until they overheat and burn. Then they put plumbing in their put0r, so the mobo now burns at 6 GHhz instead of 5 GHz. And you think AMD should build CPUs for this trivial market share?

Considering that the only time periods where AMD has been extremely profitable has been when it had the undisputed performance leadership, it would suggest that AMD should build CPUs for the enthusiast market.

But one doesn't really need to build CPUs for the enthusiast market, the market naturally gravitates to the fastest CPU for desktop and gaming applications. If you have it, then you can sell a small number of extremely profitable CPUs for $500-$1000 and a large number of profitable CPUs for $200-$300. If you don't have the performance, as AMD is, then you're stuck at selling all your CPUs at less than $200 and losing money.

 

[reb00tin]

Considering that the only time periods where AMD has been extremely profitable has been when it had the undisputed performance leadership, it would suggest that AMD should build CPUs for the enthusiast market.

But one doesn't really need to build CPUs for the enthusiast market, the market naturally gravitates to the fastest CPU for desktop and gaming applications. If you have it, then you can sell a small number of extremely profitable CPUs for $500-$1000 and a large number of profitable CPUs for $200-$300. If you don't have the performance, as AMD is, then you're stuck at selling all your CPUs at less than $200 and losing money.

Maybe the gamer market "gravitates" to the fastest CPUs. But the big money will be in portable devices, tablets and laptops. These customers never heard of computers with plumbing but they're every bit enthusiasts, that's why Trinity production is backlogged.

 

[Zarathustra[H]]

I love AMD and I hope they come back, but the reason they are no longer competitive is quite simple, and irrefutable.

AMD only was competitive with Intel on CPU performance for a very short period of time, from 1999 (launch of K7 Athlon/Duron) to ~2006 (Launch of Core 2).

What made AMD competitive in 1999 was not their own efforts, it was the following:

• Hiring of CPU engineers from Digital and using their design and know how to develop a new product.
• Licensing core DEC Alpha technologies from Digital for use in the K7 platform
• Intel's poor decisions in bringing out the failed Pentium 4 Netburst technology.

If not for the ability to take advantage of Digital technology and former employees, as well as Intel's major screwup, AMD would never have been competitive in the first place.

Once AMD was competitive with Intel, they at first seemed to approach it as the once in a lifetime opportunity it really was, using the revenue from their increased sales to first beat Intel to desktop 64bit instruction sets while intel was still messing around with that Itanium garbage, and the beat Intel to dual core CPU's.

Then they bought ATi.

They used all that cash reserve they had saved up from being a market leader in CPU performance, and dumped it into ATi, leaving not very much money to spend on CPU product development.

First we got a failed Phenom launch.

Then we got a lackluster Phenom II launch

Then we got another failure in desktop Bulldozer.

This is what happens when you don't spend enough on R&D, and AMD could no longer spend that kind of money on R&D after buying ATi.

Long term, buying ATi instead of spending the cash to remain neck at neck with Intel on the CPU front may have been the right decision to make. AMD's APU's will probably sell well, and make their way into the mobile devices of the future, where we all know the real sales volumes will be, as performance Desktops become an increasingly niche market.

I don't know for sure.

But what I do know is that they could buy ATi, or compete with Intel on x86 performance. They couldn't do both.

I also know that mid to low end power saving parts for use in mobile applications are a MUCH larger market that performance desktop CPU's, so I can't blame them for going in that direction.

And here we are.

 

[drescherjm]

This^^^