FX-57 Why Not Faster?

[(cf)Eclipse]

The idea that AMD screwed up on the Winchester I find amusing. If they did it was a damn successful screw up for sure.

same here. i actually laugh to myself every time it's suggested that winchester/venice wasn't successful for amd

 

[Erasmus354]

One question Scali...where is the Winchester with 1MB cache? Oh right there isn't one. Another question, how much cache does the high end AMD chips have? Oh right 1MB....

How much cache is San Diego going to have? Oh right 1MB. There is a reason why AMD is not producing 90nm high end parts....they dont need to or want to atm until SanDiego is released.

Do you really think that an FX-55 costs so much more than a lower clocked part with the same core? No it doesn't. The price difference between 130nm and 90nm is not as important for already high priced parts such as the FX series. The place where AMD stands to gain the most from moving to a smaller manufacturing process is in the cheaper low end parts. That is why AMD made 90nm parts for the low end before the high end, because it is more important to save money on low end parts than it is on high end parts.

Please dont use the excuse that because AMD doesn't have high clocked chips on 90nm that their 90nm process was a failure. If they felt the urge to they could have transitioned the high end parts to 90nm right away, but the fact of the matter is it simply isn't worth it to do so.

 

[mr_ouija]

There's a lot of good opinions in this thread, but I don't think all the BS was necessary. I frankly could care less if AMD is making profit or not, or that intel's 90nm process was so flawed. None of that is stuff I'm looking for in a thread about the FX-57.

That being said, I still think they're counting on the enthusiast overclocking the chip to get the most out of it. I would personally not want to buy a chip that is already at the max clock it can support. I don't care if it is 1ghz or 1thz... If I can't overclock it at all, I'm not nearly as interested in buying one. I'd be rather disappointed if you couldn't get a good overclock out of something so pricey as the FX/EE lines of processors.

 

[slamgoku]

I'd gander they leave them at a lower clock because they *ARE* the Enthusiast's chip. They expect they will be Overclocked etc, so leave them at a fraction of what they can do, and let the enthusiasts find what they are capable of. That's just my opinion though, I could be way off.


Get your [H]s here!

nevermind i KNOW the reason is money. why release a 3 ghz right away and lose all that money they could make on the 2.8 ghz fx 57 ?

 

[Scali]

One question Scali...where is the Winchester with 1MB cache? Oh right there isn't one. Another question, how much cache does the high end AMD chips have? Oh right 1MB....

Obviously more cache would make it even harder to scale up to high clockspeeds...

How much cache is San Diego going to have? Oh right 1MB. There is a reason why AMD is not producing 90nm high end parts....they dont need to or want to atm until SanDiego is released.

What kind of nonsense is that?
As I said before, 90 nm is cheaper to produce, which has many advantages...
So what reasons could you possibly give for AMD not needing or wanting these advantages? Especially considering that AMD is making losses.
And you also didn't explain why Winchester is already being replaced in favour of Venice...
It was introduced only a few months ago... You think it's a good investment to change cores every few months, with all the design, validation, and manufacturing process changes?

Please dont use the excuse that because AMD doesn't have high clocked chips on 90nm that their 90nm process was a failure. If they felt the urge to they could have transitioned the high end parts to 90nm right away, but the fact of the matter is it simply isn't worth it to do so.

It's always worth going to a lower process. You're right about budget chips being more important, but that doesn't mean higher-clocked chips are not important... Besides, we're talking about 3500+ here... I'm sure that at least some of the models above 3500+ are still quite high-volume.
And you also haven't explained why Winchester NEVER reached those high clockspeeds, and instead it is abandoned in favour of a new core... which runs hotter, yet is supposed to scale higher.

Heck, that XBitlabs article even specifically states so:
http://www.xbitlabs.com/articles/cpu/display/athlon64-venice_14.html

But the most important advantage of the Athlon 64 processors based on Venice core is the significantly higher frequency potential. This opened new horizons for further AMD processor family development, at the same time giving overclockers got new opportunities for building high-performance systems on the far not the most expensive processors.

Geez, I'm getting quite tired of these AMD fan-boys with their heads in the sand.
At least give REASONS to prove that you're right... I've given plenty of reasons for my side of the story, but all I get is crap like "AMD doesn't need to do this", or "you're wrong", without giving any good reasons.
So please, admit that you are wrong, or that you can't prove you're right... Or just prove you're right and stop harassing me with nonsense.
I have plenty of facts to back my story up anyway. So get over it.

 

[Scali]

same here. i actually laugh to myself every time it's suggested that winchester/venice wasn't successful for amd

It's a well-disguised failure (Winchester is, obviously, Venice is the fix for that failure)... Which means, the products they put out are okay (apart from the memory controller issue)... But it's about the products they DIDN'T put out.
It's just not a good move business-wise to have a core for such a short period of time. And that's where it is a failure. The core was too limited to replace the entire range of processors, so now yet another core had to be developed on the short term to fix this problem. Being available for such a short time, and in such a limited range of processors, means they didn't have much of an opportunity to redeem the R&D costs for a new core.

You can say what you want about Intel's Prescott, but it's still being used today, throughout the entire range. So it was a solid investment from Intel... The R&D costs are well-covered by now.

 

[(cf)Eclipse]

three words: stop and wait.

you know next to nothing about the new revision. i don't see how you can already call it a failure. stop this nonsense bickering
also, i didn't make any references to prescott.
read: you say your unbiased, but you're really not. now go away

 

[Scali]

you know next to nothing about the new revision. i don't see how you can already call it a failure. stop this nonsense bickering

I never said the Venice was a failure. I said the Winchester was a failure.
I even said in the previous post that Winchester was the failure and Venice was the fix.
So why do you still insist that I call Venice a failure?
You just need something to annoy me right, just because you don't like what I say... even though I am basing what I say on facts that even you cannot deny. Get over it.

also, i didn't make any references to prescott.

I did, to give an example of how far and wide a core revision usually goes... even though the Prescott is generally considered to be a failure.
I could also have taken any AMD core, from ClawHammer to Palomino, and you'd see that they all had a longer lifetime and a wider product spectrum.

read: you say your unbiased, but you're really not. now go away

It's "you're", from "you are"...
And if I am biased... prove it. Point out any biased statements that I made.
Else, stop saying crap like this about me. You are really starting to get rude and childish, and I am not enjoying it one bit.

 

[(cf)Eclipse]

my apologies. let's stop this, again. i may be biased, but it is not my intention to start any flame. this is going too far

 

[CCUABIDExORxDIE]

its been said before, itll be said again

you dare question eclipse ?!?! 100 lashes!

 

[wickedld9]

Scali, I should have figured. You seem to love taking down AMD as I've noticed over the last few months that's all that you seem to do.

Winchester was a failure eh? I'm not seeing the "obvious failure" that you claim.

There is a reason that Winchester was not produced for the higher end, it was not designed to. All of AMD's high end CPU's run 2.4GHz+ and have 1MB L2. Winchester was designed to boost the low end as a core specifically designed with low power and 512K cache. Previous to Winchester the lowest end 939 was the 3500+ at $400. It was AMD getting their feet wet in 90nm. No smart company moves their entire livelihood to something new and un-proven. ATI is a perfect example.

Hey guys! Improving something means it's a big stinking obvious failure!!
If it were a failure, we wouldn't be talking about it right now because it wouldn't be on the shelf. The 1.13GHz PIII was a failure.

Venice is the result of the learning experience that was Winchester, plus core enhancements, plus memory controller improvements, and a different manufacturing process. It is still not going to scale over 2.2GHz until the 90nm San Diego 1MB cache starts scaling up. From every person that I've seen with a Venice, it has no problems running AT LEAST 2.6GHz at default voltage, 1.4v, on stock cooling. I believe most have been able to do 2.8.

 

[lithium726]

is it so hard to think that possibly AMD released 90nm parts to the low end to hammer out issues with it before it got to the high end? taht seems to be exactly what they are doing, and nobody complains when nvidia or ATI does it, and they do it all teh time. new processes on graphics cards are always released to the lower end - with the exception of possibly the FX series (i think they went stright to 130nm there)

 

[Scali]

Scali, I should have figured. You seem to love taking down AMD as I've noticed over the last few months that's all that you seem to do.

Really now? You must have a VERY selective way of reading posts. I guess that about half of my posts don't even involve processors at all, let alone AMD ones. And even if I do mention AMD, it's not always 'taking them down'.
And what is the point of this statement of yours anyway, other than trying to discredit me, hoping to get support for what you're about to say?

For the rest, I've already given my answers, stop sounding like a broken record. And that goes for everyone in this thread.

 

[Scali]

is it so hard to think that possibly AMD released 90nm parts to the low end to hammer out issues with it before it got to the high end?

Well, if there are issues to hammer out, there must be something not entirely right in the core... which is exactly my point.
It seems that AMD went overboard with limiting the power-consumption in 90 nm, to avoid going down the same road as Intel's Prescott... And as a result the core could not scale to clockspeed even as high as the 130 nm could, so a new core was required on short notice.

taht seems to be exactly what they are doing, and nobody complains when nvidia or ATI does it, and they do it all teh time. new processes on graphics cards are always released to the lower end - with the exception of possibly the FX series (i think they went stright to 130nm there)

The difference being that these cores are generally in production for several years, with little or no changes.

 

[zeebs]

3ghz bleh. Who cares. I am not OCD enough to need 3ghz . As long as it OC's good for a fair price I don't really care if I get stuck under 3ghz.

 

[lithium726]

Well, if there are issues to hammer out, there must be something not entirely right in the core... which is exactly my point.
It seems that AMD went overboard with limiting the power-consumption in 90 nm, to avoid going down the same road as Intel's Prescott... And as a result the core could not scale to clockspeed even as high as the 130 nm could, so a new core was required on short notice.

wouldnt that be a good thing? instead of going the intel road and relesing a 130watt TDP chip becuase they need it to scale right off hte bat,they release a lower end chip while reducing power consumptoin and heat - while still running completly stable, and work on their process to perfect it for the high end. i dont see where this is a bad thing - processors are only garanteed to work at stock, and winchesters do.

 

[zeebs]

@ Scali.

Your speculation aside, there were other obvious problems with the winchester. Like the inability to run 4 DS dimms without a DDR 333 divider, and some strange quality control issues with the memory controllers on a select few chips. Could it be possible that the more powerful memory controller is pushing more heat? Yeah. I think your underlying assumption is that Venice is going to clock so much better than winchester, butfor the vast majority of chips being made theres no way to really verify that this is true. As far as we know the mem controller and new instruction set are the only thing that really make it different from winchester at all.

 

[Scali]

wouldnt that be a good thing? instead of going the intel road and relesing a 130watt TDP chip becuase they need it to scale right off hte bat,they release a lower end chip while reducing power consumptoin and heat - while still running completly stable, and work on their process to perfect it for the high end. i dont see where this is a bad thing - processors are only garanteed to work at stock, and winchesters do.

As I explained many times before already (jeez, don't you people READ!?), developing new cores is expensive. Heck, just read my earlier posts about it.

 

[lithium726]

As I explained many times before already (jeez, don't you people READ!?), developing new cores is expensive. Heck, just read my earlier posts about it.

but its not a new core. its a revision. intel does the same thing - you think the first revision of northwood could clock to 3.4ghz? hell no. AMD just decided to call teh tweaked winchester "venice"

 

[Scali]

Could it be possible that the more powerful memory controller is pushing more heat?

Not very likely, since the memory controllers are virtually identical.
I think it's their DSL-technology that allows them to scale to higher clockspeeds, but probably has the side-effect that you get more power leakage... Intel encountered similar problems with their Prescott, possibly because their strained silicon has this side-effect aswell, on 90 nm at least.

Yeah. I think your underlying assumption is that Venice is going to clock so much better than winchester

It's not an assumption, it's a fact. The fastest Winchester is 2.2 GHz, while Venice is already past that, before it's even on the market. Obviously I'm not considering overclocking results. They have nothing to do with this.
There's a reason why AMD never sold their Winchesters at more than 2.2 GHz... Perhaps the memory controller fails even more at clockspeeds over 2.2 GHz... We don't know... and AMD is not going to tell us.

 

[zeebs]

Like the other above guy said maybe they were just testing the water with winchester. Who cares how much it cost them to do it. Do I care how much AMD spends in R&D? The most important thing to me as a consumer is that they are making progressive changes.

Not very likely, since the memory controllers are virtually identical.

Umm how do you figure. Following that logic the entire freaking core is virtually identical

And what is DSL technology? Thanks.

 

[lithium726]

And what is DSL technology? Thanks.

dual stress layer. it helps venice/san diego clock higher

 

[zeebs]

Anyone have a link that they could provide me with to learn more about dual stress layer technology, or if they would be kind enough to PM me or post more information about it without going too off topic that would be cool.

 

[Mtnduey]

I'm not sure why people are expecting the Venice/SD core to perform at it's best @ stock settings before the mfg's even have an updated bios for the new cpu core. All the tests done so far have been done with the existing bios versions, the current bios versions dont really even support SSE3 do they? We dont even know if XP needs any form of an update to incorporate SSE3 for the AMD cpus do we?

So much speculation & what if's, but with the exception of those lucky review sites out there, those of us who want the new core are just drooling over our keyboards waiting for the release.

Has AMD seen tough times? Sure they have, what IT company hasn't over the past several years? Have there been issues with 90nm on both sides of the fence? Yes, Intel has had drastic problems with theirs, and AMD has had issues with cold temp rigs, overclocking @ stock voltage for some chips, and some others, but even with those issues, who's 90nm is performing better/more consistent(sp?)?

Back to the topic of the thread. FX-57 why not faster? Ace up the sleeve perhaps? Mostly it's marketing IMO. CPU's are whats making AMD money atm, so why not get the most bang for their R&D buck? I can't blame em, the FX line is a specialty market anyway, your not gonna see them in servers, you'll see em in high $$$ custom systems & custom built rigs. It takes a unique individual to pay $800 for a single CPU after all, and they know this.

Besides, who is AMD's competition right now? Intel is their only competition really, you can't count Motorola macintosh CPU's as competition since it's a market of it's own. AMD has been working with IBM's RISC proc dept for a few years now and they are making some great stides in technology.

Here where I work we're testing Opteron based servers and are liking them. I hope to be getting some into production soon myself. The CPU power of the Opteron's compared to our Intel Xeons is incredible. At lower clock speeds, lower cost & all, the AMD's are killing the Intel CPU's in our tests. It brings a smile to my face when I put together my reports

So why not make the FX57 faster? They dont have to. If they did, they would have to guarntee it's performance at that speed 24x7, and they dont want to do that, nor would I. They know people are going to OC the CPU anyway, so call it 100% stable 24x7x365 at 2.6/2.8ghz and let the end user decide how far they want to push that cpu. For the retail boxes they would have to put a heatsink in there that would keep it 100% stable too dont forget, and that means more copper, more weight, more money to ship, more cost, etc... and they want to keep costs down, bang for buck again.

Anyhow, just my 2cents worth

 

[Scali]

but its not a new core. its a revision. intel does the same thing - you think the first revision of northwood could clock to 3.4ghz? hell no. AMD just decided to call teh tweaked winchester "venice"

It is a new core... It has added SSE3, and a modified memory controller.
What you mean is called 'stepping'... That's different from new cores.

 

[Scali]

Umm how do you figure. Following that logic the entire freaking core is virtually identical

They support the same types of memory, run at the same clockspeeds, fit on the same motherboards, etc, etc.
If that still means ~17% more power increase for the entire CPU, something must be wrong with Venice... the memory controller is not the most power-hungry part on the die anyway.

 

[Scali]

I'm not sure why people are expecting the Venice/SD core to perform at it's best @ stock settings before the mfg's even have an updated bios for the new cpu core. All the tests done so far have been done with the existing bios versions, the current bios versions dont really even support SSE3 do they? We dont even know if XP needs any form of an update to incorporate SSE3 for the AMD cpus do we?

SSE3 is a set of instructions. It's independent of the bios.
And since it uses the standard SSE/SSE2 registers, it's not dependent on the OS either. Any OS supporting SSE, will automatically support SSE2/SSE3 properly... Just like MMX automatically worked with any OS that supported the FPU. As long as the registers are saved during the context switch, you have full support.
The only bios-related problem was that the CPU wasn't detected properly (wrong rating displayed and such, possibly wrong clockspeed/multiplier detected without a patch... should still operate properly if you set it manually).

 

[tsuehpsyde]

I'm still baffled as to how

A. Someone can come into a thread and more or less crap on it, and
B. Continue to crap on it, doing NOTHING with the initial topic, and
C. Doing so while running an AMD CPU in their rig?

But in a manner of speaking I guess it all ties in as the San Diego is going to be in the FX series of CPUs, which is part of the whole Venice/San Diego thing. I personally think Winchester was a test bed for AMD to try out 90nm and less power consumption....I doubt they ever intended to run winchester as their main core for a year or two. SSE3 was added more or less to keep with standard, the same way Intel has their 64-bit processing to keep up to standard with AMD. I think even without SSE3, we'd see Venice....just SSE3 was tossed in since the refresh was there anyway.

And the reason they diddn't shrink the high end CPUs was because it was a testing bed. Now that they've got it figured out for the most part, they can start moving all their products, with improved memory controller and added instructions, to 90nm as they've ironed out most of the problems. And the hit in performance for Venice? I've heard that it's linked to the fact that the fixing of the memory controller so it'll run 4x512 @ DDR400 requires it to take a performance hit (not sure why, just remember reading it). The main problem left is it can't run 1T with that much memory (last I knew) and if they don't get that fixed and release ANOTHER core with that fixed, it's a bit dumb. But as far as I'm concerned Venice is simply a revision (read: Rev. E) and SSE3 is simply tossed in the mix to stay level with Intel, and the memory controller wasn't fixed but merely changed to fit the needs of the consumer, as they've voiced an opinion they want to be able to run 4 sticks of memory at DDR400. I say changed because it gave a small but noticable hit in performance with two sticks.

And I don't know how you ever ended up buying an AMD chip if you can sit here and bad mouth them all day. And try giving your rebuttles in one reply, and save Kyle some database entries.

 

[zeebs]

So yeah, enough of this off topic core revision rambling. I am going outside.

 

[centvalny]

DFI has a new beta bios for venice or sdiego....DFI 4XX-X new beta biosAnyone with these cpu yet? Hopefully will be available today...

 

[Scali]

I'm still baffled as to how

A. Someone can come into a thread and more or less crap on it, and
B. Continue to crap on it, doing NOTHING with the initial topic, and
C. Doing so while running an AMD CPU in their rig?

A. What you call 'crapping', I call adding interesting factual info... it's the reactions to that info that caused the 'crapping' of the thread.
B. So yes, this thread is way offtopic... bit too late to do something about that now, is it... And this stuff of yours is at least as much crapping and off-topicness as before, so you should ask yourself why you bothered to post this in the first place.
C. Geez, grow up... I bought a processor, it's not like I joined some underground political movement with huge conspiracy plans or something. It baffles me how people think that everyone should praise every company that they ever bought any products from, no matter what.
I have no affiliation with either AMD or Intel, so I don't see why I should be biased towards or against either company.
Heck, even with your idea that I have to support AMD because I bought one of their CPUs, I'd be in trouble, because I also bought Intel CPUs, and Motorolas... even an IBM/Cyrix at one point.

And I don't know how you ever ended up buying an AMD chip if you can sit here and bad mouth them all day.

I bought an AMD chip because I liked its price/performance ratio. Not because I thought it was the best CPU ever made, because I obviously realized that this was not the case.
And who am I badmouthing? I am stating facts, dangit!
It is a FACT that AMD is making losses.
It is a FACT that the Winchester was only sold at 2.2 GHz at most.
It is a FACT that the Winchester is already being replaced after a few months of active service.
It is a FACT that Winchester has trouble with its memory controller and 4 banks of DDR400 memory.
It's really pathetic if you feel you have to badmouth me, just because I state some facts!
How old are you, 7?

 

[wickedld9]

Really now? You must have a VERY selective way of reading posts. I guess that about half of my posts don't even involve processors at all, let alone AMD ones.
And what is the point of this statement of yours anyway, other than trying to discredit me, hoping to get support for what you're about to say?

For the rest, I've already given my answers, stop sounding like a broken record. And that goes for everyone in this thread.

Sorry, I probably was a little harsh with that. I've just noticed in the few posts I've read there have been several heated debates and that you were right in the middle stirring the pot.

 

[centvalny]

Just stop arguing and wait for (at least) results of fx55BN 90nm which will be post here in this thread.....soon...

 

[Scali]

Sorry, I probably was a little harsh with that. I've just noticed in the few posts I've read there have been several heated debates and that you were right in the middle stirring the pot.

Let's just say that some of these AMD guys here are VERY touchy and extremely fanatic.
If you pay more attention, you'll see that I never actually say that AMD's products are bad value for money... in fact, I've even advised people to buy an Athlon on various occasions.
So in short, I am not against AMD at all. I just have a more neutral outlook on the situation. I don't believe that everything AMD touches turns into gold. Same goes for Intel, but for some reason people don't need any convincing to see that some of Intels products have some problems. In fact, in that case people seem to go overboard in the other direction... Claiming that Intels products are absolutely useless, and that you must be insane if you actually consider buying one...

Sad really, isn't it...

 

[tsuehpsyde]

I never said you had to ask them to wed and write up a pre-nup. I just find it hard to believe someone can think so poorly of a company, yet go ahead and buy their product. And as far as my replies being 'crapping' I have no idea where that's even coming from. I've never heard of someone crapping on their own thread.


Me: Hey guys, why not faster amd?

Me: AMD FRICKIN' SUCKS THEY MAKE NO MONEY!

I'm pretty sure that's not how it goes. All the replies off topic were trying to put closure to your off topic comments, but obviously that went extremely well. I'm amazed you havn't jumped in the Venice rumors thread to start dishing out your AMD realities to them, even though statements you've made have been shown as false before. You remind me of those women who never shut up, and will never admit they're wrong. Or at least as annoying as one.

Edit: And damn, if you're going to post, put an edit like this at the point of edit so I can keep track. At least you diddn't simply reply again for the 30th time to keep it more clean, but take a deep breath between replies kid, honestly. And 7? Yes, I'm 7. Good call captain factual.

 

[Scali]

I never said you had to ask them to wed and write up a pre-nup. I just find it hard to believe someone can think so poorly of a company, yet go ahead and buy their product.

I am not thinking poorly of AMD.
As I say, I recommend AMD to people... I may even buy an AMD again in the future.
You need to read what I say more closely. I am not putting AMD down, I am stating that they are in a poor financial condition, and that one of their cores was not a big success (for them, business-wise, it's fine for consumers, good price/performance).

AMD FRICKIN' SUCKS THEY MAKE NO MONEY!

If you think that this resembles what I've been saying in this thread, you have a serious reading disorder.

 

[tsuehpsyde]

lol, no, that was me being a smartass But I used that as a point that I was nowhere in here thread crapping. Thread crapping was more or less your first post when you toss in your random trivia of how AMD is on a downward spiral. And the comment about recommending AMD came after I started replying once you edited your post, hense why I diddn't address that.

You don't have to love a company to purchase their products, but I just don't see how you can have such disdain for their business practice yet recommend their products to friends/clients.

 

[Scali]

lol, no, that was me being a smartass

Yet you can't tell that I am being sarcastic when I guess your age at 7?

Thread crapping was more or less your first post when you toss in your random trivia of how AMD is on a downward spiral.

Funny, since that was actually a response, or rather a confirmation, of what others had already said previously.

You don't have to love a company to purchase their products, but I just don't see how you can have such disdain for their business practice yet recommend their products to friends/clients.

Well, their business practice is not that bad... So they seem unable to make a good profit, big deal... It's not like they have sweatshops with child labour in some 3rd world countries or so... so I don't see any moral reasons not to buy their products...
In fact, their lack of profit may be a result of their good price/performance ratio, which is exactly the reason why so many people like their products.

 

[wickedld9]

They support the same types of memory, run at the same clockspeeds, fit on the same motherboards, etc, etc.
If that still means ~17% more power increase for the entire CPU, something must be wrong with Venice... the memory controller is not the most power-hungry part on the die anyway.

I noticed that in the xbit review and I'm very curious to see how this plays out. I'm going to reserve judgment on this until there is more than 1 review. 2c is almost insignificant as the overall temp is cooler than most every video card and certainly Prescott.

 

[wickedld9]

The BIOS support problem was FUD for the most part. It came from the INQ....grain of salt.