The Bulldozer Review: AMD FX-8150 Tested
by Anand Lal Shimpi on October 12, 2011 1:27 AM ESTThe Pursuit of Clock Speed
Thus far I have pointed out that a number of resources in Bulldozer have gone down in number compared to their abundance in AMD's Phenom II architecture. Many of these tradeoffs were made in order to keep die size in check while adding new features (e.g. wider front end, larger queues/data structures, new instruction support). Everywhere from the Bulldozer front-end through the execution clusters, AMD's opportunity to increase performance depends on both efficiency and clock speed. Bulldozer has to make better use of its resources than Phenom II as well as run at higher frequencies to outperform its predecessor. As a result, a major target for Bulldozer was to be able to scale to higher clock speeds.
AMD's architects called this pursuit a low gate count per pipeline stage design. By reducing the number of gates per pipeline stage, you reduce the time spent in each stage and can increase the overall frequency of the processor. If this sounds familiar, it's because Intel used similar logic in the creation of the Pentium 4.
Where Bulldozer is different is AMD insists the design didn't aggressively pursue frequency like the P4, but rather aggressively pursued gate count reduction per stage. According to AMD, the former results in power problems while the latter is more manageable.
AMD's target for Bulldozer was a 30% higher frequency than the previous generation architecture. Unfortunately that's a fairly vague statement and I couldn't get AMD to commit to anything more pronounced, but if we look at the top-end Phenom II X6 at 3.3GHz a 30% increase in frequency would put Bulldozer at 4.3GHz.
Unfortunately 4.3GHz isn't what the top-end AMD FX CPU ships at. The best we'll get at launch is 3.6GHz, a meager 9% increase over the outgoing architecture. Turbo Core does get AMD close to those initial frequency targets, however the turbo frequencies are only typically seen for very short periods of time.
As you may remember from the Pentium 4 days, a significantly deeper pipeline can bring with it significant penalties. We have two prior examples of architectures that increased pipeline length over their predecessors: Willamette and Prescott.
Willamette doubled the pipeline length of the P6 and it was due to make up for it by the corresponding increase in clock frequency. If you do less per clock cycle, you need to throw more clock cycles at the problem to have a neutral impact on performance. Although Willamette ran at higher clock speeds than the outgoing P6 architecture, the increase in frequency was gated by process technology. It wasn't until Northwood arrived that Intel could hit the clock speeds required to truly put distance between its newest and older architectures.
Prescott lengthened the pipeline once more, this time quite significantly. Much to our surprise however, thanks to a lot of clever work on the architecture side Intel was able to keep average instructions executed per clock constant while increasing the length of the pipe. This enabled Prescott to hit higher frequencies and deliver more performance at the same time, without starting at an inherent disadvantage. Where Prescott did fall short however was in the power consumption department. Running at extremely high frequencies required very high voltages and as a result, power consumption skyrocketed.
AMD's goal with Bulldozer was to have IPC remain constant compared to its predecessor, while increasing frequency, similar to Prescott. If IPC can remain constant, any frequency increases will translate into performance advantages. AMD attempted to do this through a wider front end, larger data structures within the chip and a wider execution path through each core. In many senses it succeeded, however single threaded performance still took a hit compared to Phenom II:
At the same clock speed, Phenom II is almost 7% faster per core than Bulldozer according to our Cinebench results. This takes into account all of the aforementioned IPC improvements. Despite AMD's efforts, IPC went down.
A slight reduction in IPC however is easily made up for by an increase in operating frequency. Unfortunately, it doesn't appear that AMD was able to hit the clock targets it needed for Bulldozer this time around.
We've recently reported on Global Foundries' issues with 32nm yields. I can't help but wonder if the same type of issues that are impacting Llano today are also holding Bulldozer back.
Facebook
Twitter
RSS
430 Comments
View All Comments
saneblane - Thursday, October 13, 2011 - link
Man, you have a lot of optimism. I am a big Amd fan, but even i can remain optimistic after this mess, I mean how do you make a chip that is slow, expensive and losses to it's older brothers. Barcelona was a huge success compare to this, it only seemed bad because Expectations were high, this time around though they became higher because no one expect Amd to actually go backwards in performance. WOW that's all i can say WOWPipperox - Thursday, October 13, 2011 - link
I don't understand why you all think it's slower than its older brothers.It's not, it's faster than Thuban in practically all benchmarks...
Or do you really care about stuff like SuperPi?
Pipperox - Thursday, October 13, 2011 - link
But maybe you guys think that it's slower "clock for clock" or "core for core".It doesn't matter how you achieve performance.
What matters is the end performance.
Bulldozer architecture allows it to have higher clock speed and more *threads* than Phenom.
The penalty is single threaded performance.
Again you can't compare it to an hypothetical 8 core 4.0GHz Thuban, because they couldn't have made it (and make any money out of it).
I'll repeat, the FX-8150 is NOT an 8-core CPU.
Otherwise the i7-2600K is also an 8-core CPU... both can execute 8 threads in parallel, but each pair of threads shares execution resources.
The main difference is that Sandy Bridge can "join" all the resources of 2 threads to improve the performance of a single thread, while Bulldozer cannot.
They probably didn't do it to reduce HW complexity and allow easier scalability to more threads and higher clock speed.
Because the future (and to a large extent, the present) is heavily multithreaded, and because Bulldozer is targeted mainly at servers. (and the proof is its ridiculous cache)
bryman - Thursday, October 13, 2011 - link
how about some bios screenshots? Is there a way in the bios to disable the northbridge in the chip and use the northbridge on the motherboard? Possibly better performance, or maybe add a new ability to x-fire northbridges? (Yah imah Dreamer). imo, I dont think adding the northbridge to the cpu was a good idea especially if it pulls away from other resources on the chip, I understand what adding the northbridge to the processor does, but does it turn off the northbridge thats already on the motherboard? The northbridge on the chip makes sense for an APU but not for a perfomance CPU, why is the nothbridge even in there. I myself would rather see the northbridge on the motherboard utilizing that space intstead of the space on the cpu.If there isnt a way to turn off the northbridge on the cpu in the bios, i think the motherboard manufactures should include the ability to turn off the northbridge on the cpu. Add the ability to use the onboard northbridge in there bios, so you can atleast get bios or firmware updates to the northbridge and perhaps get more performance out of the cpu/gpu.
When the new Radeon 7000 series video cards come out, if I buy this CPU with the 6000 series northbridge in it, am I going to take a performance hit or am i going to have to buy a new processor with the 7000 series northbridge in it? or will they come out with a 7000 series motherboard that utilizes a 7000 series northbridge that turns off the 6000 series northbridge in the chip, which in turn makes it useless anyways. I myself dont like the fact if i buy this product, if i want to upgrade my northbridge/ motherboard, I might have to buy a new processor/ perhaps a new motherboard or am i just paranoid or not understanding something.
Who knows, maybe in the next couple of weeks, Mcrosoft and/or AMD will come out with a performance driver for the new processors.
If they would have come out with this processor when planned originally, it really would have kicked butt. instead we get conglimerated ideas over the five year period, which looks like the beginning idea, thrown into a 2011 die.
I am i die-hard AMD fanboy and always will be, Just kinda dissappointed, excuse my rants. I will be buying a 4 core when they hit the streets, hopefully in a couple weeks.
saneblane - Thursday, October 13, 2011 - link
From the caching issues, to the bad glofo process, to the windows scheduler, i recon that this processor wasn't ready for prime time. Amd didn't have any choice i mean they almost took an entire year extra for peet sake. Even though my i5 2500 is on it's way, am not stupid enough to believe this is the best the arch can do. Their is a good reason that interlagos cannot be bought in stores, Amd know for a fact that they cannot sell this cpu to server maker, so they are busy working on it, i expect that it might take one or even 2 more stepping to fix this processor, the multithread performance is their so they only need to get a mature 32nm process to crank up the speeds and maintain the power consumptions. IMOarjuna1 - Thursday, October 13, 2011 - link
Reviews @ other sites like toms hardware and guru 3d are starting to make this look bad. How come everyone but Anand got to review it with watercooling?? Is this site in such bad terms with AMD?B3an - Thursday, October 13, 2011 - link
Water cooling isn't magically going to help performance or power consumption in any way so why does it matter?? When you buy this CPU it comes with air cooling, and Anand was right to use that for this review.marcelormt - Thursday, October 13, 2011 - link
http://www.tomshardware.com/reviews/does-amds-athl...Patrick: The 6000+ is the fastest Athlon 64 X2 dual core processor ever, but what happened to the FX family?
Damon: Patrick, you are right. The X2 6000+ is the fastest AMD64 dual-core processor ever... so why isn't it called FX? To answer that I have to explain what FX is all about... pushing the boundaries of desktop PCs. FX-51 did that right out of the gate, with multiple advantages over other AMD processors, and a clear lead on the competition. Move forward a bit to where AMD put high-performance, native dual-core computing into a single socket with the FX-60. Fast forward again and you see FX pushing new boundaries as "4x4" delivers four high-performance cores with a direct-connect, SLI platform that is ready to be upgraded to 8 cores later this year
Ryomitomo - Thursday, October 13, 2011 - link
I'm a little surprised you only posted Win7/Win8 comparison figures for FX-8150. It would give a much complete picture if you would also post i7-2600k Win7/Win8 comparison.czerro - Thursday, October 13, 2011 - link
I think anand handled this review fine. Bulldozer is a little underwhelming, but we still don't know where the platform is going to go from here. Is everyone's memory so short term that they don't remember the rocky SandyBridge start?