Vishera Piledriver Benched

[defaultluser]

working in Excel Bulldozer is faster than thuban.

Most people who use Excel never push it to the limits where it benefits from multiple cores. You have some edge cases like extremely large spreadsheet calculations and Monte Carlo simulations, but for %99.9 of use cases out there a dual-core will give you the same performance as an octo-core.

The largest limiting factor? If you're doing ANYTHING complex with Excel, you're likely using Visual Basic For Applications. Unfortunately, VBA is single-threaded, so if you're doing anything complicated with your data involving Macros, most of your cores will sit idle.

I seriously doubt Bulldozer will have the advantage over Thuban in a test involving one of my company's multitude of VBA scripts.

 

[InternationalHat]

Most people who use Excel never push it to the limits where it benefits from multiple cores. You have some edge cases like extremely large spreadsheet calculations and Monte Carlo simulations, but for %99.9 of use cases out there a dual-core will give you the same performance as an octo-core.

The largest limiting factor? If you're doing ANYTHING complex with Excel, you're likely using Visual Basic For Applications. Unfortunately, VBA is single-threaded, so if you're doing anything complicated with your data involving Macros, most of your cores will sit idle.

I seriously doubt Bulldozer will have the advantage over Thuban in a test involving one of my company's multitude of VBA scripts.

If you're using VBA that much you should probably be looking at more appropriate statistical analysis software, mathematical software, or business intelligence and reporting tools. Excel is not the right tool if you're getting blocked and sitting around waiting on single threads and huge spreadsheet calculations.

VBA is a hack more often than not when dealing with larger data sets or stochastics. You're likely better off using mathematica or something.

If you can't get people to work in those things, I guess you can throw hardware at it, but VBA just doesn't scale in any meaningful way.

 

[defaultluser]

If you're using VBA that much you should probably be looking at more appropriate statistical analysis software, mathematical software, or business intelligence and reporting tools. Excel is not the right tool if you're getting blocked and sitting around waiting on single threads and huge spreadsheet calculations.

VBA is a hack more often than not when dealing with larger data sets or stochastics. You're likely better off using mathematica or something.

If you can't get people to work in those things, I guess you can throw hardware at it, but VBA just doesn't scale in any meaningful way.

Most of the VBA tools we have are for analyzing and formatting test data, not for simulation. I'll grant that there are better tools for the job, but Excel gets the nod because (1) it is there, (2) people are comfortable with it, and (3) we already have Excel integrated into some parts of our toolchain.

For the heavier simulations, yes we use Matlab exclusively, but like I said before %99.9 of our number crunching is analysis of results.

 

[loki_racer]

I'm sorry, but it's hard for me to take a website serious when they didn't bother to change the social media links in the sidebar of their purchased template and don't know how to properly set z-index of their main nav so it lays under the photo slideshow, but over everything else.

 

[Acer_Sheep]

pre release benchmarks and rumors are always close to truth, there was never a case where the final cpu was signifantly faster then its rumored brothers

 

[Dan_D]

pre release benchmarks and rumors are always close to truth, there was never a case where the final cpu was signifantly faster then its rumored brothers

Correct. Most of the ES silicon that hardware review sites can really get their hands on is pretty close to release hardware.

 

[gamerk2]

Sorry, I still think that the way price/performance is improving in the enthusiast segment is pathetic. I expect better than the same price after 20 months.

Where was the Q6600 priced when it was a top of the line Quad? ~$250
Where was the Q9550 priced when it was a top of the line Quad? ~$250
Where was the i7 975 priced? ~$250
Where is the i5 2500k priced? ~$250

Everything is price SEGMENTS, with older chips moving down a segment as newer, better performing chips replace them.

Hence, why a top of the line PD will cost $300-$350 at launch, regardless of performance.

 

[InternationalHat]

Where was the Q6600 priced when it was a top of the line Quad? ~$250
Where was the Q9550 priced when it was a top of the line Quad? ~$250
Where was the i7 975 priced? ~$250
Where is the i5 2500k priced? ~$250

Everything is price SEGMENTS, with older chips moving down a segment as newer, better performing chips replace them.

Hence, why a top of the line PD will cost $300-$350 at launch, regardless of performance.

Where are you getting those numbers?

q6600 release price - $851
q9550 release price - $530
the 6-core i7 970 was - $885
opposed to the i7 860 (quad) at $284

In comparison the AMD 1090T launched at $295
with the 1055T at $199.

The 955 X4 released at $245 when it was top of AMD's line.

I don't think a PD will be priced at $300, but we'll see.

 

[Skripka]

Where was the Q6600 priced when it was a top of the line Quad? ~$250
Where was the Q9550 priced when it was a top of the line Quad? ~$250
Where was the i7 975 priced? ~$250
Where is the i5 2500k priced? ~$250

Everything is price SEGMENTS, with older chips moving down a segment as newer, better performing chips replace them.

Hence, why a top of the line PD will cost $300-$350 at launch, regardless of performance.

I need to talk to whoever you're buying CPUs from...because those were not release prices. Those are prices (excluding the Sandybridge) after those parts were 2-3 generations obsolete. When 1366 was still the thing just before Sandybridge hit shelves, the i7-950 was still over $300 retail.

 

[Dan_D]

Where was the Q6600 priced when it was a top of the line Quad? ~$250
Where was the Q9550 priced when it was a top of the line Quad? ~$250
Where was the i7 975 priced? ~$250
Where is the i5 2500k priced? ~$250

Everything is price SEGMENTS, with older chips moving down a segment as newer, better performing chips replace them.

Hence, why a top of the line PD will cost $300-$350 at launch, regardless of performance.

First off, those CPUs were never top of the line. There were always Extreme Editions above them or something else. Secondly those are not their launch prices. Not even close.

Where are you getting those numbers?

q6600 release price - $851
q9550 release price - $530
the 6-core i7 970 was - $885
opposed to the i7 860 (quad) at $284

In comparison the AMD 1090T launched at $295
with the 1055T at $199.

The 955 X4 released at $245 when it was top of AMD's line.

I don't think a PD will be priced at $300, but we'll see.

Beat me to it. I got my Core 2 Quad Q6600 at around $550 when they had their first price drop.

 

[mk8]

pre release benchmarks and rumors are always close to truth, there was never a case where the final cpu was signifantly faster then its rumored brothers

I wouldn't be so sure about that in this case. The benchmarks don't seem to be faked, i don't doubt the ES is real, but if this is Vishera it wouldn't be very logical. What we know now is that the sample still has the old (tweaked?) Bulldozer cores instead of the Piledriver cores it was announced to have, and that doesn't make any sense at all. They already have the technology in Trinity, so why wouldn't they use those in Vishera?
My theory would be that the sample is not the final retail version, but just an intermediate step between Zambezi and Vishera, so any performance increase we see in the benchmarks can be attributed to the uncore improvements of Vishera, and we see another ~10% in the final version with the Piledriver cores.
I don't want to sound like a fanboy here, and i'm ready to be disappointed, but this would be the only way the whole story would add up to me, unless AMDs plan now is to have a 1,5 year architectural gap between their APUs and performance CPUs.

 

[buttons]

When the FX8350 is released, im going to build myself a badass thuban system -- 1090t with 990 fx sabertooth for like $57, i cant wait!

 

[aphasia]

When the FX8350 is released, im going to build myself a badass thuban system -- 1090t with 990 fx sabertooth for like $57, i cant wait!

why not build a thuban/990FX rig now, unless you mean to pick-up this combo on the cheap via FS thread from someone upgrading to the FX8350.

the old X6's are nice chips. they scale well to underclocking / under-volting for a low power consumption rig. properly OC'ed (3.8-4+ GHz, NB @ 2.8GHz+, RAM @ 1600MHz+), they're still a relevant high-end system for lightly threaded/heavily threaded applications.
would have been nice to see a X8 thuban on 32nm (half-way there with original Llano), course that didn't fit AMD's plans for HSA.

still a bit interested in the FX8350. if at release they out-perform my OC'ed 1100 @ 4GHz (obliviously not at stock, imagine 4.6 at least required) in almost every metric, i'd consider a 8350. if not, happy to ride out AM3+ with my old thuban.

 

[KazeoHin]

It's the Bulldozer launch all over again..... People scrambling to cling to the scraps of hope "Those pre-release benchmarks are just rumor!" "That's an engineering sample! It doesn't count" "That website is an Intel troll!"

Then HardOCP does its analysis and returns: "yep, pretty much sucks."

 

[Mr Spocko]

I'd wait to say it sucks until we get more reviews..and we know reviews do vary so do benchmarks. BD mostly sucked because it took too long, overhyped and didn't deliver the expected performance "for the price"

If they released the chips then at today's prices the reviews would have been a lot more positive. At prices right now they're quite good value processors esp if you want a decent video encoding machine at a good price.
OR if AMD had sold them as 2/3/4 core processors folks might have taken a different view ;-)

I'm not making exucses for AMD I was disappointed too, but those hefty price cuts and the cashback eased the pain a lot.

 

[buttons]

why not build a thuban/990FX rig now, unless you mean to pick-up this combo on the cheap via FS thread from someone upgrading to the FX8350.

Because im a cheapskate. a few months ago i picked up a 1045T with gigabyte 760g board for $99. I cannot believe how GOOD that is. $99 i mean a few years ago i could barely get a sempron and a cheap motherboard for $99. I didnt overclock it because i didnt trust the motherboard to handle more then 95watts, but it still worked very well for me.

 

[drescherjm]

Because im a cheapskate. a few months ago i picked up a 1045T with gigabyte 760g board for $99. I cannot believe how GOOD that is. $99 i mean a few years ago i could barely get a sempron and a cheap motherboard for $99. I didnt overclock it because i didnt trust the motherboard to handle more then 95watts, but it still worked very well for me.

I do not believe thubans are still being produced so unless you want used I expect the supply to get more scarce as time goes by so do not expect the prices to come down any further on new x6 chips.

 

[Mr Spocko]

The 1045T is still in production (limited production I would think) it has to be as you can still buy it new right now.
Stocks would have run out long ago if it were not.

Some sellers ramped up the price of the higher end PhII processors, so much so you can find stuff like a Phenom II X6 1090T for a staggering £280 odd! (pure rip off IMO) I say shove it where the sun does not shine! Same for ebay and the like overpriced rip off sellers trying to cash in on something that isn't worth the price

I don't care how much of a Thuban fan you are that is beyond a joke. Anyone with a higher end or OC'ed PhII x6 is likely not looking at FX..and should be happy enough performance wise. Anyone else well you'd have to be utterly dumb to pay a premiun for them now. The 1045T is the only logical choice if you must play with the X6 Phenom's. For me I found the FX fine for my needs.

 

[Tsumi]

The 1045T is still in production (limited production I would think) it has to be as you can still buy it new right now.
Stocks would have run out long ago if it were not.

Some sellers ramped up the price of the higher end PhII processors, so much so you can find stuff like a Phenom II X6 1090T for a staggering £280 odd! (pure rip off IMO) I say shove it where the sun does not shine! Same for ebay and the like overpriced rip off sellers trying to cash in on something that isn't worth the price

I don't care how much of a Thuban fan you are that is beyond a joke. Anyone with a higher end or OC'ed PhII x6 is likely not looking at FX..and should be happy enough performance wise. Anyone else well you'd have to be utterly dumb to pay a premiun for them now. The 1045T is the only logical choice if you must play with the X6 Phenom's. For me I found the FX fine for my needs.

Nope, just old stock that they're clearing out. AMD has completely stopped 45nm production.

 

[Mr Spocko]

Well they must have a ton of old stock that's all I can say

 

[grumnpy49x]

Everyone bought the 1055t and above processors 1st, so there is probably a large stock pile of them.

 

[octoberasian]

Nope, just old stock that they're clearing out. AMD has completely stopped 45nm production.

Well they must have a ton of old stock that's all I can say

Everyone bought the 1055t and above processors 1st, so there is probably a large stock pile of them.

Yup, pretty much any old Phenom II (X2, X4, X6), Sempron, Athlon II, and Bulldozer-based FX chips are all being sold from remaining stock at this point right now. Come 2013, they'll all be Piledriver-based processors.

Currently everything looks like this between the end of 2012 and middle of 2013:

Midrange to High-end - AMD FX (Piledriver), Socket AM3+, no on-die GPU
Entry-level and budget - Sempron (Piledriver), Socket FM2, single module/dual core and dual module/quad core, no on-die GPU
Budget to mid-range - Athlon II (Piledriver), Socket FM2, dual module/quad core, disabled GPU
Budget to mid-range - Trinity (Piledriver), Socket FM2, dual module/quad core, on-die GPU
​

Fudzilla mentioned of an FM2+ but I don't recall which processor will use it. It's Trinity-based that's for sure.

Steamroller will be the last AMD processor on AM3+ (and possibly FM2). Who knows what happens when Excavator-based processors come around? Too early to tell right now.

 

[Bdonedge]

So AMD is getting out of the high end CPU market for desktops? Did I read that correctly? Intel has a monopoly, then?

 

[Weeth]

So AMD is getting out of the high end CPU market for desktops? Did I read that correctly? Intel has a monopoly, then?

For all intents and purposes, yes. And all high end enthusiasts can now look forward to all the joys that consumers experience through monopolies: Ratcheted up prices; slowed down innovation; old products represented as new; and general b***f***ing. Thanks, AMD! That was really nice of you to just abandon the high end. NOT!

 

[Tsumi]

For all intents and purposes, yes. And all high end enthusiasts can now look forward to all the joys that consumers experience through monopolies: Ratcheted up prices; slowed down innovation; old products represented as new; and general b***f***ing. Thanks, AMD! That was really nice of you to just abandon the high end. NOT!

You seriously think they would do this if they had a choice? It's no use getting mad at them, it's not like you can design a miracle chip that's faster than Intel in every way.

And they're not pulling out. They're just shifting their focus to mobile and efficiency, which, to be honest, is where the money is. It will still be scaled up to performance parts as much as possible, but that's not the focus.

 

[tacos4me]

For all intents and purposes, yes. And all high end enthusiasts can now look forward to all the joys that consumers experience through monopolies: Ratcheted up prices; slowed down innovation; old products represented as new; and general b***f***ing. Thanks, AMD! That was really nice of you to just abandon the high end. NOT!

until intel screws up a product cycle and we all get to relive the k8 days

 

[Mr Spocko]

Intel will never have a monopoly I only buy and build AMD and in most cases I'm not in the market for "the best performance at the highest price" I will never purchase Intel processors..ever.

It's sad to see some "cave in" I'm not the line holds firm. I've dealt with AMD they're a brilliant company to work with with excellent system builder support. Sure BD could and should have been better, but heck I'm honestly pretty happy with what I paid, and what it delivers. Intel have nothing to offer in the price range for heavy threaded users like myself.

Quit wimiping out..there is no surrender around here.

 

[The Alias]

did some digging and found this : http://www.tomshardware.com/reviews/a10-5800k-a8-5600k-a6-5400k,3224-2.html
now I'm really doubting the validity of OBR now since a 5800k sans l3 has a 15% advantage in ipc in both multi and single threaded applications

 

[octoberasian]

You seriously think they would do this if they had a choice? It's no use getting mad at them, it's not like you can design a miracle chip that's faster than Intel in every way.

And they're not pulling out. They're just shifting their focus to mobile and efficiency, which, to be honest, is where the money is. It will still be scaled up to performance parts as much as possible, but that's not the focus.

Yup.

If you look at any industry right now, there is a big shift from desktop computing to mobile computing. Everyone is going mobile in the future whether we like it or not. Someone a few years ago, I believe Bill Gates or Steve Jobs or someone of that caliber, predicted that the future will be mobile and will be connected through those devices as they become more and more a part of our lives. We may soon have most of our computing done on mobile devices in the future with a small, dedicated midrange PC at home for heavier tasks. For regular consumers not in 3D rendering, video editing/encoding, or image editing and graphics design, a computer that serves their daily computing needs will make up a large share of the desktop computer market.

The sales of mobile devices is starting to or has already outpaced desktop computing. Console gaming is already seeing a slight slowdown but there are more console users than PC users now than it was maybe a decade ago.

We see trends in Intel and AMD and ARM pushing for more faster, better and efficient mobile processors and graphics chips. And on the other spectrum, server processors to handle the extra computing power for a more connected world.

The enthusiast gaming computing makes up a niche market now. It's pretty evident already. Developers are developing more games for consoles and mobile devices than having a dedicated PC game that can utilize every bell and whistle in an enthusiast PC. It's where the money lies now. Microsoft, especially, starting with Windows 8 and Windows Phone 8, are geared towards mobile devices and being interconnected to each other, not enthusiast gamers. It's the hard reality we have to face.

This is already seen in the console market as games on PCs are barely different than their console counterparts. There are so few games on the PC that outclass a console game in terms of graphics fidelity. As long as the console remains more affordable than a high-end PC with graphics barely different than a PC game nowadays, the console will gain more users. It's the difference between $200 to $300 on a console or $500 to $2000-plus for a gaming PC. Price trumps everything, and with price, consumers decide where they will open their wallets to.

A decade ago or more, consoles was something you'd ask for every Christmas or birthday for that one cool game that wasn't usually seen on a PC at that time. Consoles allowed gaming developers take advantage of the hardware without a burden of a heavy OS on top of it. Graphics in PC gaming didn't catch up to the fidelity of console games until DX10-based graphics cards and games came out sometime after PS3 and 360 consoles were released. Graphics were pretty much blocky or less detailed 3D objects when those two consoles were released. This was because console game developers can harness the bare-metal power of the underlying hardware that utilized a single hardware specification. The consoles also gave consumers with an affordable computing device for streaming movies and gaming on a widescreen TV without the need to spend hundreds or thousand-plus dollars on a dedicated gaming computer.

When I took economics in college, one of the things I've learned is that if you are a business and you specialize in only one single product or service, you will not survive. Especially in a global market and a very mobile market, you cannot just focus on one product and expect to make it to the next fiscal year. You have to expand and shift focus to where you, as a company, will have to make money.

AMD can't just rely on desktop enthusiast processors and cannot shift resources there to make them as great as Intel's. Intel outclasses AMD's net income approximately 8-to-1 as of the last fiscal year. Intel's enthusiast processors are pretty much server processors modified for desktop enthusiast PCs, hence why they cost more. Their midrange processors are the only ones in direct competition with AMD's "high end" FX processors. Intel is already on course to release its next generation Atom processors next year with on-die HD 4000 and HD 2500 graphics. They are also looking at Medfield-based mobile products. ARM is just getting into the server business or is seriously considering it.

On the other side, AMD is putting a lot of resources towards its mobile and desktop APU processor line because these may end up in future smartphones and tablets, and ultralight notebooks and prebuilt PCs. An affordable mobile or desktop computer with decent midrange graphics is very palatable for a large majority of consumers in the market. It means consumers wouldn't have to spend so much money to play games even at decent medium settings. At the other end of the computing market, AMD is investing in server technology and processors especially with the purchase of Sea Micro. They are looking at the future of cloud computing as well because everyone or nearly everyone will be connected to a cloud server of some kind. Even EA has mentioned in an article earlier that all or nearly all games from them will be connected online.

This doesn't leave a lot of choices for the desktop enthusiast crowd now. Consumers dictate where the market heads into. And, many consumers are looking at prebuilt computers they don't have to build themselves or game consoles to provide for their gaming entertainment. And, that market is larger than the enthusiast market in desktop computers. Consumers are more mobile than they were several years ago.

AMD, Intel and even Microsoft are catching up and are wanting to gain a piece of that market which is dominated by ARM, Qualcomm, Texas Instruments, Apple and Google.

Sure, neither Intel or AMD should abandon enthusiast gamers, but that's not where all the money will be in the future it seems. Smaller, powerful and more efficient processors are a necessity in this market now. AMD cannot afford to shift limited resources to combat Intel at the enthusiast high-end when that market is much less than either the mobile market and the ready-to-use computing market of average consumers, or even the server market. Not everyone is an overclocking, enthusiast gaming geek that will turn on every bell and whistle to get the best eye candy in a game. The enthusiast market is not something that any company will foolishly devote majority of their resources to when that market is very small.

Companies are being realistic. AMD is being realistic. They know they don't have Intel's resources or revenue so they will release processors that have good performance and will fit the needs of the majority of consumers out there-- not the small majority like enthusiast gamers. And, we cannot expect AMD to realistically beat Intel or match Intel at the higher midrange or high end enthusiast processors when they have neither the resources or revenue to do so. Their resources now are focused towards mobile and server markets, where most of the money will be made now.

 

[Pieter3dnow]

So AMD is getting out of the high end CPU market for desktops? Did I read that correctly? Intel has a monopoly, then?

No according to the AMD roadmap Piledriver would allow up to 15% performance over Bulldozer.

Not to sure what you are saying with the question mark?

 

[Dan_D]

So AMD is getting out of the high end CPU market for desktops? Did I read that correctly? Intel has a monopoly, then?

AMD hasn't really had a compelling high end desktop part since the Athlon X2 days. So I wouldn't they aren't getting out of that market, but rather that they've been out of it for some time.

 

[agentk7]

Interest in this review has kicked up a bit since another thread was recently created about it and the review was mentioned in another thread getting some responses.

I just thought I'd bump this thread back up since the review was pretty well dissected here already.

My own take is that I hope this review is fake or seriously flawed in some way. I'd like to believe that Piledriver will perform more like what Tom's Hardware "previewed" with their Trinity story.

 

[XacTactX]

Interest in this review has kicked up a bit since another thread was recently created about it and the review was mentioned in another thread getting some responses.

I just thought I'd bump this thread back up since the review was pretty well dissected here already.

My own take is that I hope this review is fake or seriously flawed in some way. I'd like to believe that Piledriver will perform more like what Tom's Hardware "previewed" with their Trinity story.

You know I was thinking of that too today. AnandTech benched Trinity and found that it had about 15% higher IPC at the same clocks (my memory might be off so please correct me). I don't understand how Trinity can get 15% higher IPC but Piledriver only 5% IPC.

I hope the obr-hardware review is wrong. Any competition we get in the high end sector is going to benefit us enthusiasts.

 

[buttons]

I might of been mistaken, but i thought it was 15% more performance which was a combination of clock speed and IPC, not just IPC alone.

 

[Tsumi]

It was up to 15% IPC increase, but that varies depending on the application. As I recall, applications that have very few branch mispredictions would benefit less and have a lower IPC increase than those with high branch mispredictions, since most of the work done on Piledriver was done on the front end. Up to 15% IPC increase with ~9% average is what Anandtech previewed I believe.

 

[Zarathustra[H]]

FX-8120 is 5.26 score at 3.1 GHz.
FX-8150 is 5.99 score at 3.6 GHz.
FX-8350 is 6.81 score at 4.0 GHz.

I'm hoping these improve more with final silicon.

I was really hoping for AMD to become competitive again, but in Cinebench, that 6.81 figure is half of what I am getting with my current CPU, and less than the 7.15, I was getting with my Phenom II X6 1090T overclocked to 4.1Ghz...

It's too bad, but it appears like AMD has given up in the performance desktop segment.

This may very well be the rational business thing to do, but its still sad, especially considering all the fun I had with them in the K7 days.

Oh well...

 

[Tsumi]

I'm hoping these improve more with final silicon.

I was really hoping for AMD to become competitive again, but in Cinebench, that 6.81 figure is half of what I am getting with my current CPU, and less than the 7.15, I was getting with my Phenom II X6 1090T overclocked to 4.1Ghz...

It's too bad, but it appears like AMD has given up in the performance desktop segment.

This may very well be the rational business thing to do, but its still sad, especially considering all the fun I had with them in the K7 days.

Oh well...

Well, it would really come down to how well the FX-8350 overclocks, no?

 

[Zarathustra[H]]

Well, it would really come down to how well the FX-8350 overclocks, no?

Yep, but I'm not expecting huge overclocks given how low the turbo is compared to the base clock.


They will likely be similar to Bulldozer.

Let's assume they hit ~4.6Ghz pretty reliably.

That would predict a Cinebench score of 7.83 overclocked.... With 8 cores, compared to 7.15 with 6 cores on Thuban...

 

[Tsumi]

Yep, but I'm not expecting huge overclocks given how low the turbo is compared to the base clock.


They will likely be similar to Bulldozer.

Let's assume they hit ~4.6Ghz pretty reliably.

That would predict a Cinebench score of 7.83 overclocked.... With 8 cores, compared to 7.15 with 6 cores on Thuban...

Well, there was the preview that had the FX-8350 running at 5 ghz on an H80-like cooler. If they had gotten the heat low enough to run at those speeds on what is equivalent to a high-end air cooler, then overclocking looks promising. Of course, they could have binned it as well, but it could mean that the clocks that was previously available only on watercooling for bulldozer will now be more attainable by air cooling. Bulldozer was extremely limited by the cooling solution used, unlike SB and IB, where everyone got pretty much the same clocks regardless of cooling.

Don't get me wrong, I'm not saying that Piledriver is a game changer, or that its a miracle chip. I'm just saying I don't think its going to turn out nearly as bad as you're predicting.

 

[pelo]

AMD hasn't really had a compelling high end desktop part since the Athlon X2 days. So I wouldn't they aren't getting out of that market, but rather that they've been out of it for some time.

Luckily for them, that segment of the market's role is diminishing greatly. Unfortunately, they're late in realizing that... again

BTW, those benchmarks and CPU-Z shots were from that Chinese site and were faked and varied wildly, from 30% better than BD to equal to worse. OBR just pulled the numbers from the same "trusted Chinese source."

Tom's already did a comparison of IPC between Trinity and Llano with the IPC up to 15% better. Granted, most of the gains won't be anywhere near that, it still shows the faked numbers were indeed fake. A modest 15% bump is likely, with the end on multi-threaded workloads rather than single-threaded which will probably gain very little from IPC.