Bulldozer Sandra benchmarks leaked?

Of course, these are all synthetic benchmarks. They are disappointing though.

I will be waiting for the real-world benchmarks on release day to get the other half of the puzzle.

I think this comment on that page sums everything up, and is probably the only positive comment in the entire article:
SiSoft said:
The native CPU benchmark reveals that Intel had done a very job with "Sandy Bridge" and it is hard for AMD, even with a fresh design, to keep up with its very latest. If it came earlier it could have defeated the older "Lynnfield/Nehalem" (Core gen 1).
Click to expand...
If AMD managed to atleast get Bulldozer to perform on par with Nehalem, then it might be a little bit of a disappointment, but not a total failure.:(

Keep in mind that according to some of those benchmarks, the Bulldozer doesn't even match previous gen Nehalem. Also keep in mind they benchmarked the FX-8100 2.8ghz Bulldozer pitted against a 2600K at 3.0ghz. That right there makes the comparison a little more skewed for the Intel, but that 200mhz difference doesn't account for all of the difference. It does seem extremely odd that they would downclock a 2600K to just 3.0ghz when comparing it to a 2.8ghz bulldozer. Why wouldn't they just downclock it again to 2.8ghz or use the 2600S model? :confused:
 
Meh, Im kinda tired of caring ya know? Besides, we're constantly told that we're such a small, insignificant part of the market that nobody really cares what we think anyway.
 
bullshit or not, most people are probably not expecting the bulldozer to beat the 2600k anyway.

we just want amd to catch up to within shooting range.
 
wow, is bulldozer not out yet, ROFL. What the hell has AMD been doing?

Seems really disappointing to me.
 
That article required a lot of work. As Bulldozer approaches the Intel trolls are getting desperate.
 
both cpu's had turbo enabled. the amd cpu had 3800mhz tested (2800 base/3800 turbo) vs 3600(3000base/3600turbo for the intel camp) this gave amd the clock speed advantage. its shown clearly in the spec's section. the test is valid as it was ran. they tested a lower speed bulldozer vs a lower speed sandy.

test1.jpg
 
CaptNumbNutz said:
Of course, these are all synthetic benchmarks. They are disappointing though.

I will be waiting for the real-world benchmarks on release day to get the other half of the puzzle.

I think this comment on that page sums everything up, and is probably the only positive comment in the entire article:

If AMD managed to atleast get Bulldozer to perform on par with Nehalem, then it might be a little bit of a disappointment, but not a total failure.:(

Keep in mind that according to some of those benchmarks, the Bulldozer doesn't even match previous gen Nehalem. Also keep in mind they benchmarked the FX-8100 2.8ghz Bulldozer pitted against a 2600K at 3.0ghz. That right there makes the comparison a little more skewed for the Intel, but that 200mhz difference doesn't account for all of the difference. It does seem extremely odd that they would downclock a 2600K to just 3.0ghz when comparing it to a 2.8ghz bulldozer. Why wouldn't they just downclock it again to 2.8ghz or use the 2600S model? :confused:
Click to expand...

If BD can keep up with nehalem, it's a likely buy for me. I love my first gen i5, so if I can get something just as fast per core, but with 8 of them, it's a win.
 
Question. Which Sandy Bridge are they using for the tests? I hear 2600k, but I don't see it mentioned anywhere in the article.

Also, these results are, I don't know, a bit strange. A i5 2400 is probably significantly faster than Bulldozer in that bench. How could they expect to get $300 for it?

Mind you, I know this is one bench. I wish I knew their full setup. I might have overlooked it.
 
heflys20 said:
So then it's not bullshit?
Click to expand...
it looks even and legit, however we don't know for sure that its a legit test till after launch and its confirmed from another source. the brightsodeofnews did play with a c2 es chip and wernt very impressed with it
brightsideofnews.com said:
Regarding the performance of Bulldozer, we believe it is still too early to make a final verdict. There have been numerous leaked benchmarks over the course of the last months, but in most cases the credibility of those is strongly doubted. The leaked performance numbers are allegedly from engineering samples (ES) and were lower than expected. There are differing opinions regarding this aspect. Some people think, that Bulldozer is simply that bad. Others believe, that the ES chips were hamstrung by some bug that AMD would fix in the release stepping, or the popular theory that AMD somehow managed to artificially hamper performance to keep the secret until the very last day.

Here at Bright Side of News*, we had our fair share of playing with the Engineering Samples clocked at 1.8GHz and 3.2GHz, and in both cases the performance was less than satisfactory - especially in terms of memory bandwidth, which was seriously lagging behind Sandy Bridge processors (lower efficiency).

Assuming the info from Gigabyte is valid, the stepping theory becomes less likely, as said ES chips had the same B2 stepping as is listed on the SKU table.
Click to expand...
 
heflys20 said:
Question. Which Sandy Bridge are they using for the tests? I hear 2600k, but I don't see it mentioned anywhere in the article.

Also, these results are, I don't know, a bit strange. A i5 2400 is probably significantly faster than Bulldozer in that bench. How could they expect to get $300 for it?

Mind you, I know this is one bench. I wish I knew their full setup. I might have overlooked it.
Click to expand...

Wikipedia has a few charts of all Sandy Bridge chips, and there are no 3.0/3.6 chips. The closest to those speeds is either a 2600S (2.8/3.8) or a 2400 (3.1/3.4). Wikipedia also has a listing of FX cpus, here, and the only chip with a 2.8GHz base clock is the 8100.

This benchmark raises a few questions:
1. Exactly when was the testing done
2. Exactly which 'FX' and 'Sandy Bridge' chips were used
 
heflys20 said:
Question. Which Sandy Bridge are they using for the tests? I hear 2600k, but I don't see it mentioned anywhere in the article.

Also, these results are, I don't know, a bit strange. A i5 2400 is probably significantly faster than Bulldozer in that bench. How could they expect to get $300 for it?

Mind you, I know this is one bench. I wish I knew their full setup. I might have overlooked it.
Click to expand...

Could be one of 4/8 xeons too - doesn't really matter since all of them can get downclocked to that speed.
 
I'm sure that this doesn't even pass the threshold that John was talking about, no benchmark unless it has been compiled with the correct flags.

Also the comment about OpenCL makes me wonder that every idiot on the planet can use the internet now and manage a webpage but still to slow to understand the term "open".
 
Last edited:
mrbigshot said:
both cpu's had turbo enabled. the amd cpu had 3800mhz tested (2800 base/3800 turbo) vs 3600(3000base/3600turbo for the intel camp) this gave amd the clock speed advantage. its shown clearly in the spec's section. the test is valid as it was ran. they tested a lower speed bulldozer vs a lower speed sandy.

test1.jpg
Click to expand...

not really how turbo works, since turbo usually won't push all 4 or all 8 cores to 3.8 / 3.6 ghz, usually only 1-2 that are under heavy load, if the others are not loaded.
who knows i could be wrong.
 
Digital Viper-X- said:
not really how turbo works, since turbo usually won't push all 4 or all 8 cores to 3.8 / 3.6 ghz, usually only 1-2 that are under heavy load, if the others are not loaded.
who knows i could be wrong.
Click to expand...

According to rumors AMD turbo is supposed to push all 8 cores by few hundreds MHz and few hundreds more when only 4 are utilised. SB turbo also should boost at least 100 Mhz always on 4 cores used.
 
neither amd or intel will push the full turbo to all cores under load. if they did force the chips to do it than the test would be garbage because it wouldn't show true performance.

the sandra test never showed the i7 as a 2600 model, it may have been a xeon model but most likely a pre-production chip before clock speeds were finalized.

i want to see how they do at folding@home.
 
Some of the benefits of AMD Turbo CORE include:

Up to 500MHz of additional clock speed available with all cores active. This means even with 16 cores active with server workloads, all cores can boost at the same time. For those customers that want to maximize their performance, they now have the tools to do it.
Even higher boost states available with half of the cores active. We’re not stating exactly how high processors can boost with AMD Turbo CORE, but obviously if there is room for up to 500MHz with all cores active, fewer active cores would obviously mean less power, and more headroom to recapture with AMD Turbo CORE. At launch you will see processors marketed with a base and a maximum frequency, base will reflect the actual clock speed on the processor and max will reflect the highest AMD Turbo CORE state.
AMD Turbo CORE is deterministic, governed by power draw, not temperature as other competing products are. This means that even in warmer climates you’ll be able to take advantage of that extra headroom if you choose. This helps ensure a max frequency is workload dependent, making it more consistent and repeatable.
With Bulldozer AMD now gives customers ways to maximize their processors for various roles in the data center:
Click to expand...

http://blogs.amd.com/work/2011/01/31/bulldozer-goes-to-11/

I think it's JF-AMD blog ?
 
the other testing methodology aside, they didn't test the BD chip at its default memory setting of 1866.
i understand they wanted to normalise the tests with memory at 1333 but the BD loses around 30% of its memory performance with this comparison.
 
I think these benchmarks are purely bull.

Is it really that hard to take a screen shot? Then again even if they took a screen shot it certainly would validate the numbers any because who knows what rig there were running when being tested.

While i don't expect Bulldozer to come out stomping Sandy Bridge, that would certainly be impressive. Overall i think they will bridge the gap between current amd products and Intel's sandy bridge. I would have to place it above the first gen core processors, and more efficient thanks to the 32nm process and new power saving features. It will likely not beat Sandy Bridge processors in anything that is not multi -threaded. I expect anything that can take advantage of more than 6 cores will be a victory for Amd. The 2600k can support 8 threads,but only 4 of those are actual cores the rest is a bunch of software hombo jumbo.

From the [H] perspective these chips will overclock extremely well. They have a high turbo, so at the very least they will be able to run at the turbo speed full time. I don't have any doubts about them being able to 4.5+ ghz on air cooling.

The launch day is coming soon expected launch date is Sept 19th
 
ebduncan said:
I think these benchmarks are purely bull.
Click to expand...

Looking at their website they have tests of many other cpus using the same methodology so I'm inclined to belive those tests can be real.

Bigger problem is they done them using Sandra and with turbos enabled so it's hard to tell how it translates to good old real world apples to apples benchmarks.
 
good catch on the memory speed.

also the 8150 and 8170 will both have a higher turbo frequency. (4.2 and 4.5)
 
well if that's how they test things fine so be it.

Real world apps usually follow suit from their synthetic benches. However you never know its possible the AMD turbo core did not get enabled in the bench.

I still stick my head out and say they are fake.
 
mrbigshot said:
neither amd or intel will push the full turbo to all cores under load. if they did force the chips to do it than the test would be garbage because it wouldn't show true performance.

the sandra test never showed the i7 as a 2600 model, it may have been a xeon model but most likely a pre-production chip before clock speeds were finalized.

i want to see how they do at folding@home.
Click to expand...

No core ever has 100% utilization even when OS thinks it's fully utilized running code, also certain micro ops use less power than others.

One thing I am hopeful about is AMD's decision to use TDP as a measure of Turbo headroom vs. Intel's temperature measurement. TDP is far faster to measure, which means that BD's frequency could be much more dynamic. Allowing high frequencies for busy cycles only...

Because one of the problems Intel's approach has is running turbo on NOPs or less heat generating/power demanding ops. BD should be able to react much faster and therefore increase efficiency.

I have a feeling BD's Turbo will be a pretty big part of what makes or breaks this CPU. This is why most IPC comparisons won't capture the full picture.
 
Last edited:
I hope this turbo won't be like X6 turbo where jump of voltage was so high that CPU with 3 cores could consume more than with all 6 working.
 
I don't care if it keeps up with Sandy Bridge. If it's cheaper than Sandy when it releases I'll buy one. It'll be worlds faster than my wife's Q6600 at 3.0. Hell, it'll probably be faster than my i7 920 that I can't get past 3.6 for some reason.

Price is everything for me. I like cheap performance. My Q6600 still eats up every game I throw at it so it's all win for me if the price is right.
 
Price is everything for me. I like cheap performance.
Click to expand...

It's not like the highest end BD will be $50 cheaper than the i7 2600. I would expect AMD's prices with BD to be much closer to SB prices than Phenom II was since now AMD will be competitive.
 
drescherjm said:
It's not like the highest end BD will be $50 cheaper than the i7 2600. I would expect AMD's prices with BD to be much closer to SB prices than Phenom II was since now AMD will be competitive.
Click to expand...

You're speculating that Bulldozer will be competitive with SB. No one knows for sure.
 
mrbigshot said:
the sandra test never showed the i7 as a 2600 model, it may have been a xeon model but most likely a pre-production chip before clock speeds were finalized.
Click to expand...
Since it only shows 6MB of L3 cache and 6EU on the GPU, that wouldn't make it an i7 at all, but rather something in the i5 lineup which mean no HT and HD Graphics 2000. The lowest common denominator would be an i5-2400 downclocked to 3.0Ghz, but with Turbo Boost multipliers overclocked. Basically a sub-$200 chip.

Bulldozer would need to have excellent scaling at higher clock speeds for it to be interesting as an Enthusiast Desktop part. I'd be interested to see how high Bulldozer would need to be clocked in order to match up with a i5-750 @3.2Ghz (200x16 no turbo, 1600 DDR3) with 1-4 threads, since that is the lowest I can clock my i5-750 without becoming severely bottlenecked in certain things. Though in a 2P/4P situation, Bulldozer could still be very interesting in an encoding box, if the price is right.
 
Last edited:
You must log in or register to reply here.
Back
Top