Piledriver ES benchmarks

Saw this in another thread.

That FX-8300 puts it at or near the Intel 2500K in Cinebench.

If going by the math done by a person here and another person on overclock.net, that puts the highest end FX-8350 at or near a 2600K.

The FX-8320 would then be between those two.

As a comparison to previous AMD processors, the FX-8300 is below a Phenom II 1100T, faster than the FX-8120 but slower than the FX-8150 to put it in perspective.

Moving up to the FX-8320, it should be as fast as the 1100T, faster than the FX-8120 and nearly or as fast as the FX-8150.

Finally, the FX-8350 should be faster than the FX-8150 and 1100T.

This is just going by the estimates so far that's being done by some posters here and the other forum.

It's still a very good boost in performance.
 
Meh. Decent boost...but too little too late, unless Intel meets serious delays making their new sockets.

Interesting to note that the CPU pictured in the OP of that tread has a copyright of 2011 on it...a wee bit suspicious.
 
Show me the prices! BD is on a firestorm sale right now so unless the release prices are great I think I might wait until well into next year and grab a deal on the 95w 8 core job.
 
It better have good power consumption, if not then this is once again a fail.

I don't mind higher power consumption for better performance, but if it is worse or just on par then it is useless.
 
Meh. Decent boost...but too little too late, unless Intel meets serious delays making their new sockets.

Interesting to note that the CPU pictured in the OP of that tread has a copyright of 2011 on it...a wee bit suspicious.

I've seen (and used) Westmere ES'es where the CPU says (c) Intel '08 on it (Westmere wasn't released until 2010).
 
Ill be getting one in the next year. my 1055t runs anything I throw at it so Im not worried about upgrading.
Im more in need a of a 7990 which must be lost in AMDs drawers.........
 
If that 8 core chip was 3.5ghz, it looks like it would be on par with the 6 core Phenom II's at 3.5ghz. That's still pretty sad. They made some definite improvements, but they still have a long way to go. The 8 core chip should be beating a 6 core chip by a wide margin if the IPC was as good as Phenom II.

1000
 
If that 8 core chip was 3.5ghz, it looks like it would be on par with the 6 core Phenom II's at 3.5ghz. That's still pretty sad. They made some definite improvements, but they still have a long way to go. The 8 core chip should be beating a 6 core chip by a wide margin if the IPC was as good as Phenom II.

1000

I think that 2700k is overclocked beyond 3.5GHz
 
If that 8 core chip was 3.5ghz, it looks like it would be on par with the 6 core Phenom II's at 3.5ghz. That's still pretty sad. They made some definite improvements, but they still have a long way to go. The 8 core chip should be beating a 6 core chip by a wide margin if the IPC was as good as Phenom II.

[IM G]http://www.overclock.net/content/type/61/id/992345/width/500/height/1000[/IMG]

Cinebench does not correctly read Intel clocks right...that i7 is very overclocked. Even then AMD now releasing a new architecture that only competes with Intelsat 20 month old Sandy is very meh.
 
Even then AMD now releasing a new architecture that only competes with Intelsat 20 month old Sandy is very meh.

Especially considering it isn't even released yet. AMD seems to be keeping pace an entire generation behind Intel, if those numbers are right. Not a great place to be.
 
I think that 2700k is overclocked beyond 3.5GHz
Cinebench does not correctly read Intel clocks right...that i7 is very overclocked. Even then AMD now releasing a new architecture that only competes with Intelsat 20 month old Sandy is very meh.
Not the point. Re-read my post. Compare the 6 core Thuban to the 8 core Piledriver ES. It might just be barely on par with a 6 core Thuban that has 2 less cores if it was 200mhz faster. My point is that the IPC still hasn't caught up to their previous chips. That's the saddest part of all.
 
Last edited:
If the performance is on par with Sandy/Ivy Bridge at a price similar to Sandy/Ivy Bridge, I don't see the a problem here.

Sure, they are not the fastest things on earth, I would never expect them to be. But the fact remains that they will be good for 99% of most users, this is all that matters.
 
If the performance is on par with Sandy/Ivy Bridge at a price similar to Sandy/Ivy Bridge, I don't see the a problem here.

Sure, they are not the fastest things on earth, I would never expect them to be. But the fact remains that they will be good for 99% of most users, this is all that matters.

I think what matters is the offer great value. Great value to me -- means its faster then anything intel has at that price point.

For example, I got my new 1045T six core thuban for $109 with motherboard. You can hardly buy a dual core intel for that. Thats great value...
 
I think what matters is the offer great value. Great value to me -- means its faster then anything intel has at that price point.

For example, I got my new 1045T six core thuban for $109 with motherboard. You can hardly buy a dual core intel for that. Thats great value...

That price will buy you a low end socket 1155 motherboard and a Pentium G630, which for what most people do would be about 20% faster than the 1045T...

The 1045T would crush the G630 in well threaded apps, but most people don't use well threaded apps.

Add to that, the only reason you are getting such a good price for the 1045T is because it isn't made anymore, and stores are trying to clear our their old inventory vastly discounted...
 
What seem to be AMD Piledriver FX Vishera benchmarks have popped up on the Chinese language site Coolaler. Obviously we can't vouch for the authenticity of the info but, considering this is supposedly an engineering sample, it isn't too much of a stretch to believe these could be legit. I snagged a few images for those of you that can't access the site directly.
 
It looks to be fake, Steve. The numbers there are all over the place and the date and code signify a C0 stepping.

The cinebench scores seemed legit, but some of the other stuff is way off the mark and behind Bulldozer, which isn't possible since Trinity already improved upon IPC.

Don't trust Chinese websites. Lesson learned :(
 
people too negative in here. bd sucked balls and now all the sudden people act like they expect pd to smoke intel's best for half the price or something. smells like reverse trolling imo. in what world is that a reasonable expectation? logic skills like these are why people are gullible enough to believe in global warming.

anyways, it's going to be a long hole to climb out of, that bd dug, and pd seems to be about as fantastic a start as could be expected. combine the clock speed increases with 10-15 percent ipc and you'll have chips 20 plus percent faster than bd. compare that to ivy bridge which was 3-5 percent faster than sb and obviously if that keeps up amd will surpass intel in no time on 1/100 the r&d budget, which means basically intel is the most incompetent company in the world and amd is beyond amazing.

ok the last part is hyperbole but geez, seems some reverse hyperbole is needed around here.

i was already aware since trinity benchmarks that pd would catch up to sb so this not news to me. it's still going to face a bigger deficit in gaming and general purpose benches though, so be aware of that. in gaming it will just barely catch up to the worst sb quad cores, but again, ITS A START AND A VAST IMPROVEMENT
 
Ok. The machine translation is a little rubbish. Basically this ES was sent from AMD HQ to the company he works for, and it's one of the first ES. The poster Toppc said that Piledriver supports DDR3 2400Mhz.

(These folks are from Taiwan btw, fyi)

He ran a DDR3 2520 at 1.65V, CL9,


and this is his Super PI.


The Cinebench, as a recap


3dMark 06


The L3 cache bug is still there. Here's the Vantage.


Keep in mind that the chip is likely to be a C0 revision of Bulldozer, not the Vishera. This is based on the Ext mode. being 0Fh Orichi die. We may have a reporting mistake though...
 
Last edited:
It still makes me laugh when I think of all the fanboys who were absolutely adamant that bd was going to blow Intel away.

Based on the numbers from Steve's post this looks to be another "meh" chip from AMD.
 
I really hope those screens aren't a full-on piledriver CPU, they're terrible. Here we go being all excited that AMD might make up for lost ground by beating Deneb/Thuban in IPC only to find out that NOPE. This architecture as a whole needs to go before I'll even think about buying another AMD chip. It's almost like they're not even trying on purpose just to push APUs in laptops as their primary focus now or something.
 
AMD really doesn't have anything that makes you want to buy their chips anymore do they? They'd have to sell this at $150 on the top end to make decent sales volumes, you can get a 3570K or 2500k for about $200.

I've had a lot of AMD-based systems and this just makes me sad. Oh well, I'll just enjoy my overclocked 2500k.
 
It's not an unfamiliar strategy- BD is quite obviously designed to be more effective in the rack mount space than on the gaming desktop, and the APUs on the other end should continue to smash Intel's IGPs, just the opposite of their CPUs.

AMD just needs to make their CPUs fast enough on the desktop to compete; it's almost not worth it for them to compete in the high-end desktop/workstation market where Intel rules all.
 
If the performance is on par with Sandy/Ivy Bridge at a price similar to Sandy/Ivy Bridge, I don't see the a problem here.

The problem is that it will likely only be "as fast" as SB/IB in applications that use 8 threads, and most applications don't use 8 threads. There are almost no games that use more than 4, for example, which is why Intel is crushing AMD in gaming performance right now.


But the fact remains that they will be good for 99% of most users, this is all that matters.

They are good enough for web browsing and looking at photos sure, but you could say the same about an Athlon 64 X2. Most "mainstream" users are using laptops now too, where power efficiency is more important than raw performance.
 
With the popularity of tablets the APU route might be a smart move on AMDs part.

I hope the benchies are wrong, they really need a boost to keep the market competitive.
 
You are correct, AMD might have some server products that stand out for the people that use those types of products but for the home market AMD will never catch up to Intel. It's been a good handful of years since AMD had anything remotely close in performance to Intel. So, no they really do not have anything worth buying.

Also, with all the driver issues with my 7970's not to mention performance that still falls under Nvidia. I am on my very last ATI / AMD video card product from them as well. By now, everyone knows the GTX 680 was actually a lower end card they were going to release as the 670. But because it was so powerful and largely beat AMD is most benchmarks, they decided to hold back their top end card and rename the 670 to the 680.

So Nvidia still has an ace in the hole they are no doubt still approving upon.
 
By now, everyone knows the GTX 680 was actually a lower end card they were going to release as the 670. But because it was so powerful and largely beat AMD is most benchmarks, they decided to hold back their top end card and rename the 670 to the 680.

So Nvidia still has an ace in the hole they are no doubt still approving upon.

That tin foil hat looks pretty good on you ;)
 
I know I am in the absolute minority but the current bulldozer and future pile driver cores work better for what I actually do which is extremely threaded applications and multiple instances. I know intel is faster in fact I run intel for my gaming rigs/general use but my specialized apps amd has me currently without the giant cost to bump up to server components.
 
Even if AMD pulls off a "miracle" and surpasses SB/IB all Intel has to do is crank out Haswell instead of sitting on it until they get bored.
 
Hey AMD: I have a really easy business plan for you.

Keep designing and selling fusion APUs by the millions and earning profit.

Keep designing and selling Radeon Graphics cards by the millions and earning profit.

Keep focussing you CPU R&D on high-end servers, and sell Opteron CPUs by the millions and earning profit.

take the same exact chips used for servers, sockets and all, unlock the multiplier and re-badge them as high-end desktop processors and sell them by the millions and earn profit.

There is no point spending tons of money developing yourself into a hole in the high-end enthusiast desktop market. Intel uses the same sockets and tech for their servers as they do for their desktops. You have 16-core Socket G34 processors that could wipe the floor with any Intel i7, just re-badge the heatspreader, unlock the multiplier, give 100 bucks some fresh-out-of-university graphics designer for box art, and BAM. A low-cost-invested solution for the high-end desktop market.
 
Hey AMD: I have a really easy business plan for you.

Keep designing and selling fusion APUs by the millions and earning profit.

Keep designing and selling Radeon Graphics cards by the millions and earning profit.

Keep focussing you CPU R&D on high-end servers, and sell Opteron CPUs by the millions and earning profit.

take the same exact chips used for servers, sockets and all, unlock the multiplier and re-badge them as high-end desktop processors and sell them by the millions and earn profit.

There is no point spending tons of money developing yourself into a hole in the high-end enthusiast desktop market. Intel uses the same sockets and tech for their servers as they do for their desktops. You have 16-core Socket G34 processors that could wipe the floor with any Intel i7, just re-badge the heatspreader, unlock the multiplier, give 100 bucks some fresh-out-of-university graphics designer for box art, and BAM. A low-cost-invested solution for the high-end desktop market.

its more about the platform costs and power usage then anything else. and no G34 really doesn't work in the desktop market. yes they are good chips when it comes to servers but thats a whole different ball game. a single G34 chip its self is far slower then anything intel has. its only when you get into the 2P and 4P setups where G34 is very competitive. the interlagos chips really aren't that efficient either and to get desktop clocks out of them you are talking about a 200W TDP chip.

p.s. G34 costs them nothing to develope, lol.. all they do is slap 2 zambezi chips under 1 IHS and boom you have a 8 core/16 thread interlagos. same way magny cour was done with 2 6 core istanbul chips to make a 12 core processor.

p.p.s AMD's desktop processors are server processors. its been that way since they branched off and created Opteron for the server market. there is nothing fundamentally different from the server chips and the desktop chips except that the desktop chips are clocked higher.




the thing people keep failing to realize is that AMD is not trying to compete with intel on the ultra high end market, there is no point in doing that and there is no money there for them to bother wasting the R&D resources on. what they care about is the APU and middle market(aka the average consumer) where 90% of the money is. AMD is better off letting intel waste money in the high end market while they secure their place in the APU market. intel is years behind AMD when it comes to that. the problem with enthusiast sites and i my self suffer from it as well sometimes is that we are to fixated on numbers and what effects us instead of what is actually needed in the current market. right now the problem is software and the fact that its still 5 years behind the hardware.
 
Last edited:
That tin foil hat looks pretty good on you ;)

He's actually wrong- what is the GTX680 now should have been the GTX660 Ti. Don't blame Nvidia for overpricing them, blame AMD for under-competing with GCN.

AMD's only strategy to hold until they get new silicon is to drop prices at every new Nvidia release; they have lost initiative.

Their saving grace will be getting their drivers in order for CFX. Gimme a 6GB card with 6 mini-DP ports, and another two or three 6GB cards, and five of those shiny 27" IPS panels from Korea in portrait, and call it good!
 
He's actually wrong- what is the GTX680 now should have been the GTX660 Ti. Don't blame Nvidia for overpricing them, blame AMD for under-competing with GCN.

AMD's only strategy to hold until they get new silicon is to drop prices at every new Nvidia release; they have lost initiative.

Their saving grace will be getting their drivers in order for CFX. Gimme a 6GB card with 6 mini-DP ports, and another two or three 6GB cards, and five of those shiny 27" IPS panels from Korea in portrait, and call it good!

they both did it.. its not AMD or nvidia's fault.. GK110 was so far behind schedule and had so many issues that they had no choice but to release GK104 that way. fact is it was probably better for them that it happened that way. either way if nvidia had released the GTX 680 as a 660 TI we'd still be waiting 8 months for the GTX 680. so instead they released the 660 as the 680 and now don't have to worry about rushing GK110 until Q1 of next year when AMD releases their revised GCN architecture. at the end of the day its a win win scenario for both companies.
 
Wow, that's some really underwhelming performance there. Still slower than a Phenom II. Sad.
 
If you look at the CPU-Z, These are accually Bulldozer benchmarks. It's a new stepping for Bulldozer not Piledriver.

I mean it says Zambezi in CPU-Z, and its a CO stepping

Edit: And the new stepping still sucks ass
 
A year behind is still fine with me, if it's anything like always the CPU's will be half the price of the comparable intels which in the end is what matters to most my customers.

Not sure why anyone is expecting miracles when they said they are not competing in the enthusiast arena anymore.
 
Back
Top