First bulldozer benches from AMD - surprising results (beats 980x)

[blackened23]

http://nl.hardware.info/nieuws/24619...d-fx-processor

:
In de roemruchte hotelkamer twee straten verwijderd van IDF, heeft AMD ons de eerste officiële benchmarkresultaten van de AMD FX 'Bulldozer' processors getoond. Aangezien beide tests vermoedelijk zorgvuldig zijn uitgezocht om het nieuwe platform in een goed daglicht te zetten, kunnen we op basis van deze getallen nog geen conclusies trekken. Toch willen we jullie de resultaten niet onthouden. Allereerst toonde men een vergelijking tussen een nieuwe nader gespecificeerde AMD FX processor en een niet nader gespecificeerde Intel Sandy Bridge Core i5 processor, waarbij met behulp van het programma Handbrake een videoclip van 5 minuten wordt omgezet naar H.264 video in SD-resolutie. De AMD FX processor met acht cores voert deze taak uit met gemiddeld 223 frames per seconde, de Core i5 met vier cores kwam uit op 188 fps. Beide systemen zullen volgens AMD vergelijkbaar zijn qua prijs, waarmee men wil aantonen dat AMD een betere prijs/prestatieverhouding biedt. Dat mag dan het geval zijn, maar wie er met een negatievere blik naar kijkt zal concluderen dat AMD een dubbel aantal cores nodig heeft om nog geen 20% betere prestaties te bieden. Bij de tweede demo-opstelling toonde men de game Dirt 3 in 2560x1600 resolutie, draaiend op twee Radeon HD 6790 kaarten in Crossfire. In het ene systeem waren de kaarten gecombineerd met een Intel Core i7 980X, in het tweede systeem met een AMD FX-processor. De Intel machine wist gemiddeld 80,9 fps te produceren, de AMD machine gemiddeld 82,8 fps. De veel goedkopere AMD FX-processor is sneller dan de Core i7 980X - zo wil men aantonen - al moeten we hier toch echt doorheen prikken: Dirt 3 is op deze resolutie met de gekozen videokaart zeer GPU-gelimitteerd. De 2 fps extra van het AMD-systeem kan op 101 manieren verklaard worden, die allemaal niet noodzakelijkerwijs op CPU prestaties terug te voeren zijn. Voor echte benchmarks zullen we moeten wachten totdat de AMD FX processors daadwerkelijk beschikbaar komen. Gelukkig zal dat vermoedelijk niet meer al te lang duren. Overigens toonde AMD ook een werkende laptop met Trinity-processor, de volgende generatie AMD APU, gebaseerd op Bulldozer-cores en een volgende generatie GPU. Trinity zal halverwege volgend jaar op de markt komen.

In the infamous hotel room two blocks away from IDF, AMD is our first official benchmark results of the AMD FX "Bulldozer" processor shown. Since both tests are presumably carefully chosen for the new platform to put in a good light, we can based on these numbers no conclusions. Yet we want the results do not remember.
First they showed a comparison between a new unspecified AMD FX processor and an unspecified Intel processor i5 Sandy Bridge, with the help of the program Handbrake a video of 5 minutes is converted to H.264 video in SD resolution. The AMD FX processor with eight cores perform this function with an average of 223 frames per second, the i5 with four cores came in at 188 fps. Both systems will be comparable in price according to AMD, which it wants to show that AMD a better price / performance offering. That may be the case, but who has a more negative view would conclude that AMD is looking to double the number of cores needs to less than 20% better performance available.

In the second demonstration showed up one game in three Dirt 2560x1600 resolution, running on two Radeon HD 6790 cards in Crossfire. In one system, the cards are combined with an Intel Core i7 980X, in the second system with an AMD FX processor. The Intel machine could produce an average of 80.9 fps, the AMD machine averaged 82.8 fps. The much cheaper AMD FX processor is faster than the Core i7 980X - if you want to show - though we should really stick through it: Dirt 3 to this resolution and the chosen card is very GPU thus limited. The extra 2 fps of the AMD-101 system can be explained ways, all of which are not necessarily traceable to CPU performance.

For real benchmarks, we'll have to wait for the AMD FX processors actually available. Fortunately, that probably will not take too long.

However revealed also a working laptop with AMD processor Trinity, the next generation AMD APU based on Bulldozer cores and a next generation GPU. Trinity middle of next year will reach the market.

 

[blackened23]

Dirt 3 is fairly GPU limited at that resolution, so this leaves more questions than answers. I do know that phenom barely eeked out 30-40 fps in that same test, so it appears it is a big jump from phenom.

 

[LOL_Wut_Axel]

http://nl.hardware.info/nieuws/24619...d-fx-processor




In the infamous hotel room two blocks away from IDF, AMD is our first official benchmark results of the AMD FX "Bulldozer" processor shown. Since both tests are presumably carefully chosen for the new platform to put in a good light, we can based on these numbers no conclusions. Yet we want the results do not remember.
First they showed a comparison between a new unspecified AMD FX processor and an unspecified Intel processor i5 Sandy Bridge, with the help of the program Handbrake a video of 5 minutes is converted to H.264 video in SD resolution. The AMD FX processor with eight cores perform this function with an average of 223 frames per second, the i5 with four cores came in at 188 fps. Both systems will be comparable in price according to AMD, which it wants to show that AMD a better price / performance offering. That may be the case, but who has a more negative view would conclude that AMD is looking to double the number of cores needs to less than 20% better performance available.

In the second demonstration showed up one game in three Dirt 2560x1600 resolution, running on two Radeon HD 6790 cards in Crossfire. In one system, the cards are combined with an Intel Core i7 980X, in the second system with an AMD FX processor. The Intel machine could produce an average of 80.9 fps, the AMD machine averaged 82.8 fps. The much cheaper AMD FX processor is faster than the Core i7 980X - if you want to show - though we should really stick through it: Dirt 3 to this resolution and the chosen card is very GPU thus limited. The extra 2 fps of the AMD-101 system can be explained ways, all of which are not necessarily traceable to CPU performance.

For real benchmarks, we'll have to wait for the AMD FX processors actually available. Fortunately, that probably will not take too long.

However revealed also a working laptop with AMD processor Trinity, the next generation AMD APU based on Bulldozer cores and a next generation GPU. Trinity middle of next year will reach the market.

It's not faster than a Core i7-980X. Just no. However, I CAN see it [FX-8150] matching or beating just by a slight amount the 2600K in multi-threaded. In single-threaded it'll get its ass handed to it, though.

Looks like AMD still needs to figure out what's causing the low Cinebench scores and solve it before the retail chips ship for revenue. Even then, given that AMD needs eight cores to compete with Intel's four, it means they're probably only be good for multi-threaded applications.

The chip that looks the most interesting is the FX-8120. They're all unlocked, so I'm sure you could get similar OCs in comparison to the 8150. Plus it should 'only' be $230.

 

[Idontcare]

Dirt 3 is fairly GPU limited at that resolution, so this leaves more questions than answers. I do know that phenom barely eeked out 30-40 fps in that same test, so it appears it is a big jump from phenom.

It speaks to platform capability, which will be limited by the slowest components.

AMD is showing they've eliminated the platform limitation that was intrinsic to Phenom-based systems with the new bulldozer platform.

That's a good thing.

 

[blackened23]

It's not faster than a Core i7 980X. Just no. However, I CAN see it [FX-8150] matching or beating just by a slight amount the 2600K in multi-threaded. In single-threaded it'll get its ass handed to it, though.

Looks like AMD still needs to figure out what's causing the low Cinebench scores and solve it before the retail chips ship for revenue. Even then, given that AMD needs eight cores to compete with Intel's four, it means they're probably only be good for multi-threaded applications.

The chip that looks the most interesting is the FX-8120. They're all unlocked, so I'm sure you could get similar OCs in comparison to the 8150. Plus it should 'only' be $230.

Do you have something against AMD? Understand that all benchmarks floating around the web at the moment are probably early samples or falsified.

For me, these results leave more questions than answers. But I do know that a CPU getting 5 in cinebench probably *would not* get 83 fps in dirt 3 at 2560. But i'm eagerly awaiting official results 🙂

For me , i'm keeping an open mind. If bulldozer is a great product, that will be good for the community even if you stick with intel products.

 

[Silverforce11]

CF scales near 100% in dirt 3, depends on the AA settings it may or it may not be GPU limited. With 8xAA, 2x 6970 gets around 80 fps.

 

[BlueBlazer]

Dirt 3 is fairly GPU limited at that resolution, so this leaves more questions than answers. I do know that phenom barely eeked out 30-40 fps in that same test, so it appears it is a big jump from phenom.

Your title is misleading actually (should be titled "AMD FX first official showing and benchmarks"). At those resolutions, GPU limitations starts kicking in, AMD FX is only faster by 2.9 fps (can find many examples of this 1 to 5 fps differences in GPU limited situations on different platforms and motherboards). 😉

Meanwhile the Handbrake results are at best dismal (if you compare against Core i7 2600K and Core i 980X). Its already been posted and discussed here. 😉

Handbrake results (based on this reference)....

Base result: Core i5 at 188 fps, AMD FX at 223 fps

For Core i5 2500K to Core i7 2600K >> (188 * 21.6) / 16.32 = 249 fps
For Core i5 2500K to Phenom II X6 1100T >> (188 * 19.35) / 16.32 = 224.6 fps

In both cases Core i7 2600K and Phenom II X6 1100T are still faster than "AMD FX" :hmm:

 

[Idontcare]

Your title is misleading actually (should be titled "AMD FX first official showing and benchmarks"). At those resolutions, GPU limitations starts kicking in, AMD FX is only faster by 2.9 fps (can find many examples of this 1 to 5 fps differences in GPU limited situations on different platforms and motherboards). 😉

Meanwhile the Handbrake results are at best dismal (if you compare against Core i7 2600K and Core i 980X). Its already been posted and discussed here. 😉

Handbrake results (based on this reference)....

Base result: Core i5 at 188 fps, AMD FX at 223 fps

For Core i5 2500K to Core i7 2600K >> (188 * 21.6) / 16.32 = 249 fps
For Core i5 2500K to Phenom II X6 1100T >> (188 * 19.35) / 16.32 = 224.6 fps

In both cases Core i7 2600K and Phenom II X6 1100T are still faster than "AMD FX" :hmm:

BlueBlazer, wouldn't we expect the 2600K to be higher performing based on price?

 

[Riek]

Your title is misleading actually (should be titled "AMD FX first official showing and benchmarks"). At those resolutions, GPU limitations starts kicking in, AMD FX is only faster by 2.9 fps (can find many examples of this 1 to 5 fps differences in GPU limited situations on different platforms and motherboards). 😉

Meanwhile the Handbrake results are at best dismal (if you compare against Core i7 2600K and Core i 980X). Its already been posted and discussed here. 😉

Handbrake results (based on this reference)....

Base result: Core i5 at 188 fps, AMD FX at 223 fps

For Core i5 2500K to Core i7 2600K >> (188 * 21.6) / 16.32 = 249 fps
For Core i5 2500K to Phenom II X6 1100T >> (188 * 19.35) / 16.32 = 224.6 fps

In both cases Core i7 2600K and Phenom II X6 1100T are still faster than "AMD FX" :hmm:

Those numbers seem wrong you use. A result with 12times less fps can't be used to measure a performance delta. This means that the test was completely different and results are not interchangeable.

Same as Sb-E results which were a mere 15% faster.

So without the exact settings and scenarios you can't extrapolate it to other cpu's with HT.

 

[thilanliyan]

For real benchmarks, we'll have to wait for the AMD FX processors actually available. Fortunately, that probably will not take too long.

Your thread title says "First Bulldozer benches..."? 😕

 

[Silverforce11]

BlueBlazer, wouldn't we expect the 2600K to be higher performing based on price?

Not shit. This guy missed the point. The FX CPU benched here is similar to the i5 pricing and beats it handily.

 

[frozentundra123456]

Do you have something against AMD? Understand that all benchmarks floating around the web at the moment are probably early samples or falsified.

For me, these results leave more questions than answers. But I do know that a CPU getting 5 in cinebench probably *would not* get 83 fps in dirt 3 at 2560. But i'm eagerly awaiting official results 🙂

For me , i'm keeping an open mind. If bulldozer is a great product, that will be good for the community even if you stick with intel products.

I dont understand why you think Axel has something against AMD. From the limited information that we have now, his post seem spot on to me.

 

[Abwx]

Your title is misleading actually (should be titled "AMD FX first official showing and benchmarks"). At those resolutions, GPU limitations starts kicking in, AMD FX is only faster by 2.9 fps (can find many examples of this 1 to 5 fps differences in GPU limited situations on different platforms and motherboards). 😉

Granted it s GPU limited , it will be limited the same way for both
plateforms , no ?....

You are aware that 990X has more mem channels , so it s likely
that BD compensate with better throughput..

 

[BlueBlazer]

BlueBlazer, wouldn't we expect the 2600K to be higher performing based on price?

I'm comparing between Core i7 2600K and Phenom II X6. As can be seen, AMD FX looks to be positioned against Core i5 range. 😉

Those numbers seem wrong you use. A result with 12times less fps can't be used to measure a performance delta. This means that the test was completely different and results are not interchangeable.

Same as Sb-E results which were a mere 15% faster.

So without the exact settings and scenarios you can't extrapolate it to other cpu's with HT.

If you take those results as scaling by cores/threads (just look at the delta between the 4-core Core i5 and 8-"core" AMD FX in that article), then you have to find a comparable benchmark results which scales with cores/threads. Thus I picked the Guru3D's as reference as it shows this characteristic. The other Handbrake results looks borked, IMHO (including the Toms Hardware one). :hmm:

Not shit. This guy missed the point. The FX CPU benched here is similar to the i5 pricing and beats it handily.

Quote from my other post here.....

I have also noted that "Both systems will be comparable in price". That is if you also take into account the (cheaper) price of AMD boards.

 

[VirtualLarry]

Granted it s GPU limited , it will be limited the same way for both
plateforms , no ?....

You are aware that 990X has more mem channels , so it s likely
that BD compensate with better throughput..

Or this could speak to a PCI-E Crossfire limitation somewhere, on the Intel platform. Perhaps they used a board with 16x/8x/8x on Intel, and used the two 8x slots, and on the AMD platform, they had two 16x/16x slots.

"If all you have is lemons - make lemonade!"

 

[sm625]

It's not faster than a Core i7-980X. Just no.

Do you have something against AMD?

Just no.

 

[blackened23]

Or this could speak to a PCI-E Crossfire limitation somewhere, on the Intel platform. Perhaps they used a board with 16x/8x/8x on Intel, and used the two 8x slots, and on the AMD platform, they had two 16x/16x slots.

"If all you have is lemons - make lemonade!"

There's really no performance difference between dual x8 and dual x16 - 1% at best. I'd think intel would have the upper hand platform wise, this is all speculation on my part though.

 

[Voo]

Granted it s GPU limited , it will be limited the same way for both
plateforms , no ?....

We're talking about a 2.3% performance delta here with a test that is probably at best accurate to what? 1%? That doesn't seem especially significant.

 

[Riek]

I'm comparing between Core i7 2600K and Phenom II X6. As can be seen, AMD FX looks to be positioned against Core i5 range. 😉

If you take those results as scaling by cores/threads (just look at the delta between the 4-core Core i5 and 8-"core" AMD FX in that article), then you have to find a comparable benchmark results which scales with cores/threads. Thus I picked the Guru3D's as reference as it shows this characteristic. The other Handbrake results looks borked, IMHO (including the Toms Hardware one). :hmm:

So everything is borked that doesn't prove your point? Thats at least a borked way to look at things. Given the results of Guru3D they did something extremely heavy, which is not the case for the conversion done by AMD. So results are not exchangeable. In contrast with the other results which was less intensive and reached similar high fps rates. This also gives an indication that running test A on software ABC can give different results than test B.

We're talking about a 2.3% performance delta here with a test that is probably at best accurate to what? 1%? That doesn't seem especially significant.

Indeed those scores are completely irrelevant. It does show the platform isn't limiting gpu performance. Which is at least good to know for people want to use dual gpu configurations.

 

[Borealis7]

I Don't know what to think anymore!

 

[AtenRa]

I Don't know what to think anymore!

just wait for the official release date reviews 😉

 

[BlueBlazer]

Or this could speak to a PCI-E Crossfire limitation somewhere, on the Intel platform. Perhaps they used a board with 16x/8x/8x on Intel, and used the two 8x slots, and on the AMD platform, they had two 16x/16x slots.

"If all you have is lemons - make lemonade!"

Here are some quotes (source)....

Then the 2nd test was most likely the FX-8150 vs the i7-980x which we all know loses to the i7 2600k in gaming at stock settings. The FX gets a 2-3% win at a crazy high resolution on a game that is most likely cherry picked.

And here....

It seems there's a subtle reason they have been demonstrating Dirt only all along.

Anyway, one of Dirt's development partner is AMD. 😉

 

[IndyColtsFan]

I Don't know what to think anymore!

I do -- I gave up and ordered a 2600K build last night. I don't believe it will beat the 980 or the 2600K in anything except maybe heavily multi-threaded apps. I could be wrong, but all the signs point to disappointing performance and I wasn't going to wait yet another month and then have them possibly delay again.

EDIT: As BlueBlazer mentions above, if these benchmarks are, in fact, accurate and we discount a GPU bottleneck (which is probably happening), this is a cherry-picked benchmark.

 

[BlueBlazer]

So everything is borked that doesn't prove your point? Thats at least a borked way to look at things. Given the results of Guru3D they did something extremely heavy, which is not the case for the conversion done by AMD. So results are not exchangeable. In contrast with the other results which was less intensive and reached similar high fps rates. This also gives an indication that running test A on software ABC can give different results than test B.

Whatever benchmark conditions and test data they used, it has to scale nicely with more cores/threads to show AMD FX 8-"cores" advantage. And that type of setup, would require all cores almost or fully loaded otherwise it will have similar fps or lower results (for the CPU with lesser IPC). Example of another borked result here (Core i7 980X loses to Core i5 2500K! 😱)

 

[Ferzerp]

Oh yay, a video card benchmark.

With it being a video card benchmark, you really need to do more than a single run on each, get the standard deviation from runs and figure out if that 1.9 fps difference is even statistically significant before you could even begin to say anything about processor performance.

But hey, cheers on the video card benchmark.