Vishera Piledriver Benched

I found the most interesting thing about that was the "Funny Slide" part. I am perplexed at why AMD would bother with that crap.

Edit: To me on one hand the improvement in multithreaded applications looks good however how much is that a result of the higher stock speed?
 
Last edited:
Well, since i only care about x264 encoding, it's not THAT bad. But i imagine for all the other gamers here, it will be disappointment. I am also curious to see power consumption. I wish there was a 95W version! Then i 'd surely get it.

Pricing will now be crucial. If they insist on unrealistic prices, i think i will keep my 1090t.
 
Well, since i only care about x264 encoding, it's not THAT bad. But i imagine for all the other gamers here, it will be disappointment. I am also curious to see power consumption. I wish there was a 95W version! Then i 'd surely get it.

Pricing will now be crucial. If they insist on unrealistic prices, i think i will keep my 1090t.

agreed.
was hoping to replace my 1100, the cinebench 11.5 result on the 8350 are about the same as my x6 @ 4Ghz. the single threaded test is still worse (8350: 1.07 vs 1100: 1.20 - @4GHz).

if the bench tests is to be believed, either the MB bios or CPU needs a bit more work as the original tester said the system has stability issues.

i think come release time, the factors that will decide if i purchase is price, OC'ing headroom & power consumption.
 
I'll wait for more serious reviews to turn up. To much cursing and swearing in that to be taken too seriously. Also Eng sample not final retail processor.

You can't expect a review to be taken seriously if you just blast off like some knuckle head..he should learn to write in a more considered manner.
 
agreed.
was hoping to replace my 1100, the cinebench 11.5 result on the 8350 are about the same as my x6 @ 4Ghz. the single threaded test is still worse (8350: 1.07 vs 1100: 1.20 - @4GHz).

if the bench tests is to be believed, either the MB bios or CPU needs a bit more work as the original tester said the system has stability issues.

i think come release time, the factors that will decide if i purchase is price, OC'ing headroom & power consumption.

If this benches are the real thing, then i wonder what they have been doing all this time.
They haven't even touched this arc at all. They just tweaked it and overclocked it, that's it.
Rory said "Its enough" Go figure that one out. I would of fired him by now.
This reflects to the server parts as well, its not only for us gamers. It sucks.
My 1090T smokes this Vishera piledriver at 4,1 Ghz for sure. And it can go well beyond that.
 
I found the most interesting thing about that was the "Funny Slide" part. I am perplexed at why AMD would bother with that crap.

Edit: To me on one hand the improvement in multithreaded applications looks good however how much is that a result of the higher stock speed?

Great question friend. Look at the Cinebench numbers. http://2.bp.blogspot.com/-SHo71RNBND0/UEM8lNmAbVI/AAAAAAAACzo/1FEb9r3rUPQ/s1600/graf2.jpg

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.

Going from the 8120 -> 8150, a 16% increase in clock speed increases the score by about 14%. Going from the 8150 -> 8350, an 11% increase in clock speed increases the score by about 14%.

In Cinebench, Piledriver seems to only have about 5% higher IPC than Bulldozer.
 
regardless of the bench results, at least it's a taste.
the AMD forum has been starved of anything interesting to talk about for a while now.

if the 8350 proves to be a decent clocker, it may alleviate the average single thread performance. did read the use of clock-mesh tech did limited the frequency of the CPU a bit. guess we'll just have to wait on retail samples.
 
If it's real, the MT performance looks good. Single threaded performance is still abysmal. It performs like an older server chip in that way, where relatively slow 1T doesn't matter very much, but it could be a problem (again) on the desktop.

Regardless, as obr noted at the end, it'll be priced where it competes, likely much below $250. The FX label doesn't really count for much anymore, at least not in price or competitive performance.
 
Unfortunately i think there is no dedicated steamroller based desktop CPU.
Only APU.

There is a non-APU Steamroller, but AMD is planning on releasing it one year AFTER the APU code-named Kaveri is released. Kaveri APU is out in 2013 with the FX/server variants a year later in 2014.

It's going to be a long wait.

If that is the trend AMD will follow, then Excavator-based processors and APUs won't be out until 2015 (APU) and 2016 (FX/Opteron). (That's a "big if" though. If you read some articles online, AMD may be moving towards having an APU-only lineup from APU to FX to Opteron chips. Again, that's all rumors right now and nothing set in stone.)
 
If you read some articles online, AMD may be moving towards having an APU-only lineup from APU to FX to Opteron chips. Again, that's all rumors right now and nothing set in stone.)

Well i have some slides i have uploaded, check them out. These are original slides from AMD: http://www.2shared.com/document/_8ub9F--/AFDS_2012_Keynotes_Mark_Paperm.html

Yes i think AMD is planing to have APU-only lineup in all the markets.
Those slides are quite clear i think to what AMD is planning to do.

Road_Map.jpg
 
Last edited:
I predict the overall response will be "wait for steamroller" ;-)

It's the Bulldozer story all over again. High expectations --> first benchmarks ---> denial-disbelief ---> official release ---> acceptance of reality and grief ---> wait for... Oh, wait, no more AM3+ CPUs!

Wait for... or "hurry to Intel!".

I myself intend to stay for a good while with AM3+ and as far as i know, Piledriver will be the last compatible CPU.

For those who won't be happy with Piledriver, biting the bullet and going with Intel seems the only way and the sooner the better.
 
Guys, what in the world are we going to do when AMD gets out of the performance desktop space? Right now AMD doesn't have a competitive part and Intel's prices have stayed almost the same for 20 months (for i5 and i7). I really hope something happens to shake the high end out of the lull it is in.
 
Yup... It's called Haswell....:D

Keep dreaming :p Haswell is another GPU-focused architecture. With the exception of the ISAs, we're seeing another 10% performance bump.

I don't understand the sentiment, though. Who in their right mind doesn't have enough CPU power? If you feel you don't then there's the 2011 socket option there. For people who are running workstations then they don't mind splurging 1.5k on a platform. Likewise, neither do serious gamers who spend over a grand on their GPUs and another grand on monitors anyway.

If Intel and AMD kept pandering to the small and insatiable minority (read us) they'd have gone out of business years ago.
 
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.
 
Keep dreaming :p Haswell is another GPU-focused architecture. With the exception of the ISAs, we're seeing another 10% performance bump.

I don't understand the sentiment, though. Who in their right mind doesn't have enough CPU power? If you feel you don't then there's the 2011 socket option there. For people who are running workstations then they don't mind splurging 1.5k on a platform. Likewise, neither do serious gamers who spend over a grand on their GPUs and another grand on monitors anyway.

If Intel and AMD kept pandering to the small and insatiable minority (read us) they'd have gone out of business years ago.

People who do serious number crunching. 2011 isn't that much of a step up. That is why F@H folks skip 2011 entirely and use 4p g34 servers...because for the dollar and Watt, 2011 is pretty sucky in performance.
 
People who do serious number crunching. 2011 isn't that much of a step up. That is why F@H folks skip 2011 entirely and use 4p g34 servers...because for the dollar and Watt, 2011 is pretty sucky in performance.

I don't disagree, but Intel didn't introduce the platform for the average consumer. The 2011 platform is just the server mobo-and-chips with higher frequencies and cheaper price tag minus some features. If you're looking for the best performing chips then 1155 isn't for you. They do incredibly well for an overwhelming majority of desktop users and the fact that they do speaks volumes to the workloads (and unoptimized software) that we encounter on a daily basis. If a laptop chip with a useless on-die GPU can get such great acclaim then it shows just why Intel and AMD aren't bothering catering to the enthusiast.

AMD is smart in pulling out of the desktop race. It sucks for prices, sure, but they're in the business of making money and not appeasing the unappeasable.
 
I don't disagree, but Intel didn't introduce the platform for the average consumer. The 2011 platform is just the server mobo-and-chips with higher frequencies and cheaper price tag minus some features. If you're looking for the best performing chips then 1155 isn't for you. They do incredibly well for an overwhelming majority of desktop users and the fact that they do speaks volumes to the workloads (and unoptimized software) that we encounter on a daily basis. If a laptop chip with a useless on-die GPU can get such great acclaim then it shows just why Intel and AMD aren't bothering catering to the enthusiast.

AMD is smart in pulling out of the desktop race. It sucks for prices, sure, but they're in the business of making money and not appeasing the unappeasable.

Even 1155 chips are great buys for price/performance in number crunching. They aren't the best performing but from a bang for the buck measurement they have been a homerun. There's a reason why many enthusiast type folks and "Normals" have 1155 and not 2011. The "enthusiast" level of Intel's lineup ended up not catching on that much IMHO unlike how many folks had 1366.

Whether it is smart for AMD to give up on the performance desktop front has yet to be seen. Could end up biting them, as Intel has far more money to throw at whatever they please. The performance tier may not move that many parts, but it is the basis for the popular reputation of a product line. When your FX "enthusiast" chips have a rep for running hot, drinking power, being not that fast, nor that cheap...people don't look at the rest of the lineup. Offhand, other than their APU lineup, I don't know what the other chips in AMD's lineup are after the high-end FX ones...Back when Phenom 1/2 was around and the thing I knew the Athlons off hand.
 
They've attempted to put desktop performance ahead of mobile but it's only gotten them stuck in a deeper ditch. Then they create an APU and they sell over 30 million chips

Even 1155 chips are great buys for price/performance in number crunching. They aren't the best performing but from a bang for the buck measurement they have been a homerun

That's my point, though. Considering just how well a mobile-first chip can do in the desktop with a huge portion of it's die dedicated to the GPU and is still able to perform so well is a clear sign that massive increases in performance aren't required. If Haswell brings only a 10% bump people will still be here bragging about synthetic benchmarks and nothing to show for it in real world applications and scenarios just like a transition to 2011 wouldn't amount to any tangible benefit either.

It makes sense to just make a very good APU that can handle the laptop and desktop markets. There's no need to stake your reputation on IPC and single-threaded performance when it's only there for a select market that doesn't make the company enough money. I'd much rather have current Trinity core performance with a 7750-like GPU than 2x Sandy performance with no on-die GPU. Alternatively, I'd take lower power consumption above either of those.While most readers here wouldn't agree with that, the overwhelming majority of the market does. Both Intel and AMD pander to those who pay the most, not those who scream the loudest.

Also, it's the software that has to catch up and not the hardware. You'd get a 50% increase in certain benchmarks and performance in select applications just by a recompiling for new ISAs yet developers still don't give a shit. Why should they? People are too busy raving about a 10% bump in synthetic benchmarks and don't know nor care about actual real world applications because even a 50% difference isn't worth the amount of work required (and in many cases it's very minimal) to either party.
 
Last edited:
There is an argument that says average Joe doesn't even need much GPU power so an integrated GPU solution is effecitvely wasted on such buyers. I've built plenty of pc's with integrated onboard Mobo graphics to know that's mostly corrrect.

I think it's open to debate, but people are probably reading too much into it. I see it as not much more than the Intel on CPU GPU's ie use it or not up to you, AMD are merely following that line IMO that is. They might do it a bit better (should do esp the ATI purchase and experience)

I don't believe it automatically means no competition to Intel. If FX ends up as an APU it will be a sideshow as AMD cannot afford to abandon the higher margins on higher priced CPU's
 
Nothing wrong with all AMD cpu's going to have graphics. Don't use the gpu if you don't want to. Intel has i5 and i7 cpu's with the onboard gpu.Maybe AMD will also have a model that has it disabled if you want.
 
Nothing wrong with all AMD cpu's going to have graphics. Don't use the gpu if you don't want to. Intel has i5 and i7 cpu's with the onboard gpu.Maybe AMD will also have a model that has it disabled if you want.

The big problem is TDP. Perhaps if AM4 was a 150W to 175W socket, AMD could include an APU and 8 or 10 cores and still clock the cores at 4+ GHz.
 
Pretty sure Amd plans to go to all APU chips not because they are getting out of the the desktop market. But With Steamroller Core and HSA it makes sense to have on die graphics x87.

Bulldozer was launched with this dream of combining computer resources to compute your work loads. Amd has been pushing open cl. When you stand back and look at the big picture Amd is moving towards a goal. Its not like Bulldozer is a complete fail. It has good integer performance, while it lacks the FPU side of things. Now if you look at the GCN gpu architectural it is focused on compute. Or x87 fpu compute to be more specific. Now image them working together, well you have a super fast combination. Cpu that handles the integer loads and a Gpu that handles the Fpu loads.

Anyways speculation but its not hard to see what Amd is trying to do. Maybe its a bit native as it takes years and years for support of new architecture to make it to software. Heck even some things these days are still single threaded.

Anyways to be on topic, these benchmarks can be taken with a grain of salt. Guy clearly isn't very serious about his benchmarks or writing for that matter. More likely to be a Intel Troll than anything else. Not saying these benchmarks are fake by any means. Most reviews are not biased to a certain company, and always try and point out the positives, while also making people aware of the negatives. This guy simply says this thing sucks. Yet for all intensive purposes these chips could cost 100$, then it would be great. Just an example not a fact.

I don't really care about Pile drivers Stock Performance. I want to see how it overclocks.
 
I don't really care about Pile drivers Stock Performance. I want to see how it overclocks.

What is the point seeing how it overclocks? The performance is not there as we have seen in the benchmark. If you just want an under performing cpu just to clock it to gain some speed and to consume much more power, i cant see the logic in that.
This thing is not for me. Maybe for others its ok, i don't know. :confused:
 
WOW AMD gives up.... .........................................................................................

Just amazing..........................

Wonder how much of this is economics?
 
I don't see them giving up no idea where that came from. Folks are going to look at the overclocking esp on the FX8 series which does get pretty heavy power drain wise for big ish overclocks. If the power drain is less, the newer ones will OC better and we hope at least pull less power doing it.

But we will have to wait and see.

I don't expect a big uplift with PD (not a bad one but not huge), I do expect steamroller to be a good step forward, die size shrink..should help all round, higher clocks, AMD are working on the predictions, and they should OC better with the new process/die shrink. I would hope SR is AM3+..if it is some might wait for this.

Unless PD offers at least 10%+ improvement I can't see many updating if they are runing a BD.
 
I don't see them giving up no idea where that came from.

AMD are working on the predictions, and they should OC better with the new process/die shrink. I would hope SR is AM3+..if it is some might wait for this.

There is no indication that they will have a dedicated SR desktop CPU, Nor do they have any road maps for that. They are already introducing APU's in the server market so i think they are done with dedicated desktop CPU's.

I guess we will have to wait and see until it is official.
 
markpaper;1039103120 The performance is not there as we have seen in the benchmark. If you just want an under performing cpu just to clock it to gain some speed and to consume much more power, i cant see the logic in that.

What benchmarks? These from some random Chinese guy nobody knows? Also please provide some proof on power consumption for Vishera since you seem to know it's power consumption when overclocking.
 
What benchmarks? These from some random Chinese guy nobody knows? Also please provide some proof on power consumption for Vishera since you seem to know it's power consumption when overclocking.

So now i am getting attacked?
The link i found and provided here at HF is as it is. You are free to make whatever you want out of these benches.
These benches are not mine. "If it was my preview", i wouldn't of given a link, i would just post the results in here directly.
 
So now i am getting attacked?
The link i found and provided here at HF is as it is. You are free to make whatever you want out of these benches.
These benches are not mine. "If it was my preview", i wouldn't of given a link, i would just post the results in here directly.

You seemed to take these results as proof. Also you talk of power draw when overclocking when there is no numbers yet on power draw.
 
You seemed to take these results as proof. Also you talk of power draw when overclocking when there is no numbers yet on power draw.

Agreed! I think you guys need to chill out and wait for the final product benches and OC results.
 
You seemed to take these results as proof. Also you talk of power draw when overclocking when there is no numbers yet on power draw.

Nobody in here complained about something, you seem to be the first one.
Well aren't we all waiting to see how it performs when overclocked?
If this benches are real and according to the die shots that have been taken its quite safe to say that power draw would be pretty much the same thing as bulldozer.
It could be improved but not by much. I have speculated, i fail to see where is the issue here.
 
My 8120 is at 4.5 by choice. Can go higher. It doesn't draw alot of power really unless I use all 8 cores in something like Intel burn in test.The temp in that gets to 58c and stays no matter how long I run it.In all my games temp runs around 52c. Playing games its pretty dam good at 4.5. Im hoping the piledriver clocks a tad higher and I would be happy with the same wattage draw.
edit..... btw it folds every night. I shut it down when I go to work as I only have a PC power & cooling silencer 750 thats around 5 years old that powers it and the msi gtx 580. I get worried about how long it will last.
 
...
Anyways to be on topic, these benchmarks can be taken with a grain of salt. Guy clearly isn't very serious about his benchmarks or writing for that matter. More likely to be a Intel Troll than anything else. Not saying these benchmarks are fake by any means. Most reviews are not biased to a certain company, and always try and point out the positives, while also making people aware of the negatives. This guy simply says this thing sucks. Yet for all intensive purposes these chips could cost 100$, then it would be great. .....

This is the same guy that posted leaked benches of BD ES, and basically said the same thing: IT SUCKED ASS!!! Everybody jumped on him for being a troll and trying to smear AMD's name... Well, a few days later (more like a week) official benches came out, and guess what (?) EVERYBODY named BD a steaming turd... So, I wouldn't totally disregard his leaks... To be honest, everything that's derived from BD microarchitecture, sans a MAJOR redesign, is going to be a steaming pile of dung... With Steamroller, AMD may have a slight chance of reaching Sandy Bridge's performance, but will most likely be plagued by the same outrageous power characteristics that have held BD back so much. Hell, if not the ridiculous TDP, I'd go to Microcenter TOMORROW, and pick up an 8120 + mobo combo just to play with...
 
This is the same guy that posted leaked benches of BD ES, and basically said the same thing: IT SUCKED ASS!!! Everybody jumped on him for being a troll and trying to smear AMD's name...

In that situation, didn't OBR benchmark the CPU? These benchmarks are from somebody in China that has a "final ES" Piledriver CPU.

EDIT: I see that this one is OR-C0 now. Hmm... Well, I'll still wait til it is released.
 
Back
Top