Cache and Memory Performance

I mentioned earlier that cache latencies are higher in order to accommodate the larger caches (8MB L2 + 8MB L3) as well as the high frequency design. We turned to our old friend cachemem to measure these latencies in clocks:

Cache/Memory Latency Comparison
  L1 L2 L3 Main Memory
AMD FX-8150 (3.6GHz) 4 21 65 195
AMD Phenom II X4 975 BE (3.6GHz) 3 15 59 182
AMD Phenom II X6 1100T (3.3GHz) 3 14 55 157
Intel Core i5 2500K (3.3GHz) 4 11 25 148

Cache latencies are up significantly across the board, which is to be expected given the increase in pipeline depth as well as cache size. But is Bulldozer able to overcome the increase through higher clocks? To find out we have to convert latency in clocks to latency in nanoseconds:

Memory Latency

We disable turbo in order to get predictable clock speeds, which lets us accurately calculate memory latency in ns. The FX-8150 at 3.6GHz has a longer trip down memory lane than its predecessor, also at 3.6GHz. The higher latency caches play a role in this as they are necessary to help drive AMD's frequency up. What happens if we turn turbo on and peg the FX-8150 at 3.9GHz? Memory latency goes down. Bulldozer still isn't able to get to main memory as quickly as Sandy Bridge, but thanks to Turbo Core it's able to do so better than the outgoing Phenom II.

L3 Cache Latency

L3 access latency is effectively a wash compared to the Phenom II thanks to the higher clock speeds enabled by Turbo Core. Latencies haven't really improved though, and Bulldozer has a long way to go before it reaches Sandy Bridge access latencies.

The Impact of Bulldozer's Pipeline Windows 7 Application Performance
Comments Locked

430 Comments

View All Comments

  • TekDemon - Wednesday, October 12, 2011 - link

    Yeah I paid $179 for my i5 2500K and it hums along at 4.8Ghz (can hit 5Ghz+ but I wanted to keep the voltages reasonable). Clock for clock bulldozer is slower since it's only competitive when the higher clocked part is compared to a stock 2500K.
  • jleach1 - Friday, October 21, 2011 - link

    Their cores offer, what 75% the speed of a normal core?

    The fact is, this supposed "8" core processor performs worse than AMDs own 6 core processor. There's no way we can get away with calling it an 8 OR a 6 core.

    For all intents and purposes, it's a quad core.
  • estarkey7 - Wednesday, October 12, 2011 - link

    You took the words right out of my mouth! I am a big AMD fanboy, and I was waiting with baited breath to jump on the bulldozer bandwagon for my next rig (and I probably still will). But this is ridiculous! I'm a computer engineer and where the hell were the simulations AMD? Seems like you could have halved the L3 and kept in the extra FP resources and been better than what you are doing now.

    Also, don't bitch about that Windows 7 doesn't realize the architecture of Bulldozer, you knew that 18 months ago, so you should have been writing a patch so that would have been a non issue.

    The absolutely, positively only reason i will by an 8150-FX is that my current desktop is a dual core Athlon running at 2.2GHz. So to me, the performance increase over my current desktop would be massive. But on second thought, if I have stuck with such a slow system this long, I might another 3-5 months for Piledriver.
  • Taft12 - Wednesday, October 12, 2011 - link

    <i>The power consumption is absolutely through the roof -- unacceptable for 32nm, really!</i>

    Uhh, you did see the bar graph for idle power usage, right? And keep in mind this is an 8-core CPU compared to 4- and 6-core competitors.

    Like you, I'm also very interested in the 4- and 6-core Bulldozers. Anand let us down by only reviewing the flagship Llano. Hopefully he doesn't do the same with Bulldozer.
  • Tom Womack - Wednesday, October 12, 2011 - link

    Yes, the idle power is significantly worse than either of the Sandy Bridge platforms he's comparing it to
  • JasperJanssen - Wednesday, October 12, 2011 - link

    What Anand reviews is mostly down to what AMD will let him have -- even sites the size of Anandtech don't simply get to call and order parts from a catalogue for review samples.
  • Taft12 - Wednesday, October 12, 2011 - link

    AMD doesn't have much control over "review samples" that can be purchased at retail, as you can do with the A4-3300 et al. for weeks now
  • enterco - Wednesday, October 12, 2011 - link

    I read that 'at 1920x1200/1080 the gaming performance depends much mure on the GPU. Anyway, I'm happy with my i5-2500k ;-), Bulldozer does not seem to worth the wait.
  • ninjaquick - Wednesday, October 12, 2011 - link

    Blame shitty game developers.
  • AssBall - Wednesday, October 12, 2011 - link

    Kinda what I was thinking. When they are all developing games for a 6 year old 3 core PowerPC system with 512MB RAM (xbox) instead of a computer, its no bloody wonder.

Log in

Don't have an account? Sign up now