Fudzilla: Bulldozer performance figures are in

[Arkadrel]

Thuban - 346 mm2
Bulldozer - 315mm2
Sandy Bridge - 216mm2

It is definitely smaller than the Phenom II X6 but SB is 99mm2 smaller and a portion of the SB die is dedicated to an IGP. If it barely matches the Core i7 2600K in terms of performance then it's not good news for BD. D:

Yes however its likely that the Bulldozer 8core version will be faster than the 2600k with applications that make use of 8 threads.

Think of it another way:

4 core bulldozer ~ 315/2 = ~157mm22
4 core Sandy Bridge = ~ 216mm2

Will the 4 core bulldozer be faster than the 4 core sandy bridge?
Probably not, but it ll sell cheaper (because its abit smaller die size it should work).

Lets say Bulldozer 4c has 85% of Sandy Bridge's 4c performance.
well (1 - 157/216) x 100% = ~27.4% smaller size, means they can probably compete (price wise).

and offer same price/performance to sandy bridge.


This is assumeing they make a "native" 4c bulldozer version (and not just harvest bad dies from the 8core versions).

 

[Riek]

Yes however its likely that the Bulldozer 8core version will be faster than the 2600k with applications that make use of 8 threads.

Think of it another way:

4 core bulldozer ~ 315/2 = ~157mm22
4 core Sandy Bridge = ~ 216mm2

Will the 4 core bulldozer be faster than the 4 core sandy bridge?
Probably not, but it ll sell cheaper (because its abit smaller die size it should work).

If they would make a 4core BD mask they would probably be able to compete with a 4core (non HT) SB. Given the rumoured clockspeeds of 3.6 and higher for the FX4 series, which can get better with a smaller die and better placed cache interconnects. (but i don't think they would have different masks for their lineup)

If i understand TB correctly from the slides (max turbo when a module is in c6)
FX4 will run at 3.9GHz for 2 threads (rumoured speed 3.6 -3.9)
FX6 will run 3.9GHz for 4threads (rumoured speed 3.3 to 3.9)
FX8 (125W) will run 4.2 for 6threads (rumoured speed 3.6 to 4.2).

So given the clock advantage BD has in all areas, i would indeed assume they can be competitive, even for the FX4. Compared to the i5 2300 FX4 would have a 800MHz(25% -> 28%0 advantage in 1-4 threads and 500MHz compared to the i5 2400 (15.5% -> 16%)

 

[lau808]

315/2 wont = 157 just because its 4 cores. still need room for all the other items, mem controller and w/e else is on there.

and .... what's a mask???

 

[frostedflakes]

and .... what's a mask???

http://en.wikipedia.org/wiki/Photomask

 

[Abwx]

Given the die size, it should be close to 50% better than a PII X6 ,
or else AMD engineers are unable to do basic maths comparisons..

Why should they prefer to work on a CPU that is 45% bigger than
an hypothetic PII X8 while having, as propogated ad nauseam
by a few trolls , less IPC and so on.?.

.

 

[dma0991]

Yes however its likely that the Bulldozer 8core version will be faster than the 2600k with applications that make use of 8 threads.

Think of it another way:

4 core bulldozer ~ 315/2 = ~157mm22
4 core Sandy Bridge = ~ 216mm2

Will the 4 core bulldozer be faster than the 4 core sandy bridge?
Probably not, but it ll sell cheaper (because its abit smaller die size it should work).

Lets say Bulldozer 4c has 85% of Sandy Bridge's 4c performance.
well (1 - 157/216) x 100% = ~27.4% smaller size, means they can probably compete (price wise).

and offer same price/performance to sandy bridge.


This is assumeing they make a "native" 4c bulldozer version (and not just harvest bad dies from the 8core versions).

I was dumpster diving through my pile of BD rumors and this image popped out. That was a pretty good guess for how big an 8 core BD would be and an additional 5mm2 to the supposed leaked images by S|A. If it is also true then a single SB core would have almost the same die area as a BD module and that would mean that a 2 BD module/4 cores would be not be comparable against a 4 core SB. My assumption is that a 4 core BD might only win against a 2 core SB.

I think a 4 core BD would be more likely to be a faulty processor that is binned from an 8 core BD. If in the event that a 4 core BD does offer the same price/performance as a 4 core SB, that would mean that an 8 core BD would totally destroy the Core i7 2600K by a huge margin which I doubt is what is going to happen. :\

 

[blckgrffn]

eh? you can use an AM3 PHII/AII in most AM2+ motherboards as the AM3 CPUs have DDR2 memory controllers in them as well as DDR3. There should be no need to track down a 940 or a 920.

Most being the operative terms - I have AM2+ mobo's that don't support AM3 processors myself - and several which do.

Also, it was expected that AM2+ boards were going to be able to support AM3 in some fashion. Do we have any word on AM3+ supporting FM2? Is it a different pinout? If so, it is a very different situation than AM2+ to AM3.

 

[Riek]

I was dumpster diving through my pile of BD rumors and this image popped out. That was a pretty good guess for how big an 8 core BD would be and an additional 5mm2 to the supposed leaked images by S|A. If it is also true then a single SB core would have almost the same die area as a BD module and that would mean that a 2 BD module/4 cores would be not be comparable against a 4 core SB. My assumption is that a 4 core BD might only win against a 2 core SB.

I think a 4 core BD would be more likely to be a faulty processor that is binned from an 8 core BD. If in the event that a 4 core BD does offer the same price/performance as a 4 core SB, that would mean that an 8 core BD would totally destroy the Core i7 2600K by a huge margin which I doubt is what is going to happen. :\

Can you specificy what you mean with an SB core and a BD core?
I don't see a default 4core SB. There are 4core SB running at 2.8GHz->3.1GHz and 3.4GHz->3.8GHz. there are 4cores without HT and with HT. All make a huge difference in the performance spectrum.

According to the rumoured clocks:
1-2 threaded wise the rumoured FX linup differs 7.5% from FX4 to FX8!!!
1-2 threaded wise the current SB lineup differs 22.5% from 2300 to 2600

3-4 threaded wise the rumoured FX lineup differs 16.6% from FX4 to FX8
3-4 threaded wise the current SB lineup differs 21% from 2300 to 2600

In this case an FX4 running at 3.6GHz TB to 3.9GHz might very well be competitive to an 4core SB... in case of the i5 2300 it has 800MHz advantage in everything.
I give the FX4 alot more chance on dominating at least one quad core SB then i give FX8 to dominate the i7 in 1-4threads.

 

[dma0991]

Can you specificy what you mean with an SB core and a BD core?

Since all in the FX range is overclockable I would put it up against an overclockable SB processors like the Core i5 2500K and Core i7 2600K.

In this case an FX4 running at 3.6GHz TB to 3.9GHz might very well be competitive to an 4core SB... in case of the i5 2300 it has 800MHz advantage in everything.
I give the FX4 alot more chance on dominating at least one quad core SB then i give FX8 to dominate the i7 in 1-4threads.

Having a FX4 dominating the lower end quad core SB like the Core i5 2300 and Core i5 2400 is plausible. A stock comparison between FX8 vs Core i7 2600K might end with the FX8 better in multithreaded performance.

If you consider that higher clocks give an advantage, then we have already seen how well SB can overclock. As of now we have no idea how well BD will overclock and I'd like to see the performance results of a side by side comparison between a 5.7GHz Core i7 2600K vs >4.2GHz(?) FX-8150.

 

[bryanW1995]

5.7 is an extreme outlier on a 2600k, 4.7 would be a much better comparison to use.

 

[inf64]

Check this out:
http://diybbs.zol.com.cn/11/11_100430.html <- supposedly 4C 3.9Ghz Zambezi (with power limit removed?)



Same thread page six:
http://diybbs.zol.com.cn/11/11_100430_6.html
8C 2.8Ghz ,A1 revision?? So the samples seem to work fine(bug-wise) but have been power limited to stay within certain power cap.

Quote from the guy who posted this:

8-core results here

Bulldozer and then check the performance of eight threads can not pay would in 9000, the main reason for the test problems is power supply and motherboard BIOS issue, bulldozers high demand for power, and now the BIOS only 60-70&#37; of shipments performance only.

The logic above is the following: first silicon revisions (A1,B0,B1,B2?) were functioning performance-wise (mostly) as simulations predicted BUT they were very leaky and were gobbling power like crazy.So in order to validate the platform,AMD has used the new feature in Bulldozer design called "power cap" in order to limit the power draw of the CPU and still make it work on latest 900 and 800 series AM3+ boards (check the official Bulldozer blog about this useful feature in Interlagos variant ). The effect of this was that those early Bulldozer samples were throttling down aggressively (voltage and frequency wise) in order to stay within the spec. This resulted in much lower clocks then what applications like CPU-z reported- in the range of 60-70% of specification.Turbo functionality was impaired also.
After the platform is validated, AMD was already producing the B3(C0?) which supposedly has the leakage/power draw problems fixed and clock target was being met within the 95/125W specs( say 3.6Ghz 8150 8C is now hitting all targeted specifications,Turbo included,all within 125W -unlike the early A1/B0/B1 which were gobbling power like crazy with poorer yields). So after all things said,retail will perform 40-60% better than what the latest leak from chiphell showed and what the two images above (kinda) illustrate-if they are genuine that is.

As for the numbers for both 4C and 8C they kinda confirm it's possible. 4C 3.9Ghz allegedly gets 18.9K,8C 2.8Ghz allegedly gets 23K. Scaling is probably not perfect going from 4C to 8C ( i assume around 1.7x,software and hardware limitations):
18.9K / 3.9 x 2.8 x 1.7 =23K. So the scores kinda align. But the problem is that scores are roughly 1.9x higher per core than Phenom... I can't see from where this speedup comes. So take with salt

 

[Idontcare]

The logic above is the following: first silicon revisions (A1,B0,B1,B2?) were functioning performance-wise (mostly) as simulations predicted BUT they were very leaky and were gobbling power like crazy.So in order to validate the platform,AMD has used the new feature in Bulldozer design called "power cap" in order to limit the power draw of the CPU and still make it work on latest 900 and 800 series AM3+ boards (check the official Bulldozer blog about this useful feature in Interlagos variant ). The effect of this was that those early Bulldozer samples were throttling down aggressively (voltage and frequency wise) in order to stay within the spec. This resulted in much lower clocks then what applications like CPU-z reported- in the range of 60-70% of specification.Turbo functionality was impaired also.
After the platform is validated, AMD was already producing the B3(C0?) which supposedly has the leakage/power draw problems fixed and clock target was being met within the 95/125W specs( say 3.6Ghz 8150 8C is now hitting all targeted specifications,Turbo included,all within 125W -unlike the early A1/B0/B1 which were gobbling power like crazy with poorer yields). So after all things said,retail will perform 40-60% better than what the latest leak from chiphell showed and what the two images above (kinda) illustrate-if they are genuine that is.

As for the numbers for both 4C and 8C they kinda confirm it's possible. 4C 3.9Ghz allegedly gets 18.9K,8C 2.8Ghz allegedly gets 23K. Scaling is probably not perfect going from 4C to 8C ( i assume around 1.7x,software and hardware limitations):
18.9K / 3.9 x 2.8 x 1.7 =23K. So the scores kinda align. But the problem is that scores are roughly 1.9x higher per core than Phenom... I can't see from where this speedup comes. So take with salt

I'd believe it.

We all saw the 100W+ power-consumption of a lowly 2.9GHz quad-core llano chip, the power numbers for glofo's 32nm process seem to be inordinately high compared to where it needs to be for the TDP tiers.

It could all be BS but it is a very plausible scenario.

 

[tincart]

Check this out:
http://diybbs.zol.com.cn/11/11_100430.html <- supposedly 4C 3.9Ghz Zambezi (with power limit removed?)

This would put a 4C Bulldozer on par with a 2600k @ 5.0 GHz for this particular benchmark: http://2dayblog.com/2010/11/23/sand...awesome-performance-with-5ghz-in-air-cooling/

 

[Accord99]

If Bulldozer had so much performance with 4 cores, AMD wouldn't even need to bother with 8 cores for the desktop. And there would have been no reason to delay it due to power use because they could have used lower clocks/voltage while still fielding a far more competitive and smaller product than Thuban.

 

[Gundark]

18.9K / 3.9 x 2.8 x 1.7 =23K. So the scores kinda align. But the problem is that scores are roughly 1.9x higher per core than Phenom... I can't see from where this speedup comes. So take with salt

In my speculations several things came to my mind.

1. Phenom have decoder advantage over Intel ( could be wrong on that one ) but they are ill fed becouse of inferior branch predictor and could not make that advantage.
2. AMD with BD will (hopefully) have branch predictor comparable to Nehalem. Known improvements for AMD so far:
http://www.anandtech.com/show/3863/amd-discloses-bobcat-bulldozer-architectures-at-hot-chips-2010/6

Quote: In Phenom II, the branch prediction and instruction fetch logic are run in lockstep - when one stalls, the other also stalls. Branches are predicted as they are encountered. If the fetch logic grabs an x86 branch instruction, the prediction logic works in parallel to predict the likely target of that branch. However if the branch is incorrectly predicted, subsequent branches aren&#8217;t predicted until the current mispredict is correctly resolved. As a result, the fetch logic and prefetchers can&#8217;t work and potential performance is lost.
In Bulldozer the branch prediction and fetch logic are decoupled. The predictor now produces a queue of future fetch addresses. Even if there&#8217;s a mispredict the branch predictor can continue to fill its prediction queue with targets. The fetch logic can then check this queue of addresses against what&#8217;s in the instruction cache to avoid future misses in L1.
With Phenom AMD implemented comparable prefetching logic to what Intel did with Core. In Bulldozer, AMD is ramping up the aggressiveness of those prefetchers. There are independent prefetchers at both the L1 and L2 levels that support larger numbers of strides and large stride sizes (both compared to what exists in current AMD architectures). There&#8217;s also a non-strided data prefetcher that looks at correlated cache misses and uses that data to prefetch into the caches.

3. Now, best case scenario. With improved branch predictor, decoders are now well fed and can make advantage over Intel.

4. It is known that Intel compilers have hindered AMD CPUs. With BD that is not the case anymore.

If AMD menaged to pull this out, it's very small possibility that it can acctualy outperform Intels IPC ( Yes, I'm dreaming ).

 

[Abwx]

Yet another "benchmark".....

 

[guskline]

Yet another "benchmark".....

Doesn't shock me as to the 2500k since it's a 4 core 2500k vs an 8 core FX-8150. I am surprised that its that much higher than a 2600k 4core/8 hyperthreading. Where did this benchmark come from?

 

[BlueBlazer]

Check this out:
http://diybbs.zol.com.cn/11/11_100430.html <- supposedly 4C 3.9Ghz Zambezi (with power limit removed?)



Same thread page six:
http://diybbs.zol.com.cn/11/11_100430_6.html
8C 2.8Ghz ,A1 revision?? So the samples seem to work fine(bug-wise) but have been power limited to stay within certain power cap.

This guy was exposed as a fraud. Original thread here >> http://tieba.baidu.com/p/1181059002?pn=1

His previous fraudulent CPUZ screenshot >> http://tieba.baidu.com/p/1174976541?pn=9

Photochopping evidence >> http://tieba.baidu.com/p/1181059002?pn=6 and http://tieba.baidu.com/p/1181059002?pn=7

 

[busydude]

"benchmark".....

LOL. Pretty easy to do that in EXCEL.

 

[BlueBlazer]

Doesn't shock me as to the 2500k since it's a 4 core 2500k vs an 8 core FX-8150. I am surprised that its that much higher than a 2600k 4core/8 hyperthreading. Where did this benchmark come from?

From here >> http://obrovsky.blogspot.com/2011/08/bulldy-615.html

Don't get "punk'd" again.

 

[guskline]

my 2500k at 4400 scored at 3400

 

[RussianSensation]

Trucrypt benchmarks? Seriously? "Whoaaaaaaa, 2600k is a beast. It's 5x faster than i7-975X. Must get one!"

 

[drizek]

Trucrypt benchmarks? Seriously? "Whoaaaaaaa, 2600k is a beast. It's 5x faster than i7-975X. Must get one!"

Honestly, AES-NI is one of the features I'm most looking forwards to with Bulldozer. I don't encrypt now because I don't want to sacrifice performance.

 

[RussianSensation]

Honestly, AES-NI is one of the features I'm most looking forwards to with Bulldozer. I don't encrypt now because I don't want to sacrifice performance.

Ok but how many times are you going to do this task? Once you to encrypt your hard drive / partition and then the rest is easily managed in real time, right? 2500k gets 2.5+ GB/sec. So a 3TB hard drive will be finished in less than 30 minutes if my math is correct? Even if you have 30TBs, that's less than 5 hours of encryption. And after that? You are seriously telling me it's worth spending $300 for that 1 task over modern processors? Even if it took you 2 full nights (16 hours), that's still not a big deal, unless I am missing something?

 

[frostedflakes]

The stuff done on the fly is still pretty CPU intensive. AES hardware acceleration offers a pretty solid boost in performance for those who use encryption. Plus it frees up CPU cycles for other stuff.

http://www.anandtech.com/show/2901/5