The Bulldozer Review: AMD FX-8150 Tested
by Anand Lal Shimpi on October 12, 2011 1:27 AM ESTCache 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:
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 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.
Facebook
Twitter
RSS
430 Comments
View All Comments
silverblue - Friday, October 14, 2011 - link
We don't know the server performance yet. There's also some frantic investigations going on into why the client benchmarks were so poor, so watch this space.IceDread - Friday, October 14, 2011 - link
I always observe. I hope AMD makes a comeback because the market needs competition to work in favor of the customer.Analytics believes Amd is becoming irrelevant on the processor market, which is really bad. They believe AMDs share is worth 4 dollar each now and not five dollar which they believed before.
I do not know if new drivers would help Amd out or not but 10% more or less does not cut it. I really hope Amd makes a comeback with a new cpu in a couple of years but only time will tell. How they could release this one is not understandable, they had to known the values.. something is not always better than nothing..
chrone - Wednesday, October 12, 2011 - link
even phenom ii x6 is way better. :(piroroadkill - Wednesday, October 12, 2011 - link
Because there is no longer anything to wait for. Bulldozer is an absolute failure. Oh dear.That's not helpful at all, because a competitive AMD CPU is just what the market needs right now.
MossySF - Wednesday, October 12, 2011 - link
The worse part about this review is the MB diagram. Earlier this year, the diagrams for the SB950 showed a 4GB/s Alink III Expression connection. The diagram in this review shows 2GB/s -- the same as the previous SB850 and competing Intel chipsets.Who cares about CPU speed ... they're all close enough. We need I/O speed way more. I have a new server with5 x 6Gbps Sandforce SSDs in a RAID0 array and we hit the 2GB/s limit with just 2 drives. (2GB/s bidirectional is 1GB/s each way.)
And no, I have zero confidence that any 4-port/8-port RAID controller has enough power either. Maybe the most expensive 24-port ones can do it but I am not going to continually buy and return $2000 controllers until I find one that is beefy enough. Especially since the majority of the RAID functionality is completely wasted with SSDs.
nexox - Wednesday, October 12, 2011 - link
OT, but you may want to try a SAS2 HBA based on the LSI 2008 chip - They're generally around $150 and I promise they can do:A) Far higher performance than any crappy motherboard controller.
B) Way more than 2GB/s full duplex.
If you're worried about storage performance while using an on-board disk controller, you're just going about it all wrong, especially if you think you're going to gain much using their crappy software raid.
gvaley - Wednesday, October 12, 2011 - link
This monster, in this particular test, adds over 144W going from idle to full load! (For comparison the i2600K adds mere 78W and performs a notch better.) Assuming it already wastes ~10W at idle and even by factoring in the increase in power usage coming from the chipset/memory, I still very much doubt it that this...hm..thing...can fit into the stated 125W TDP.Good thing Anand didn't do the performance per watt maths...it would've painted a devastating picture.
f4phantom2500 - Wednesday, October 12, 2011 - link
i think i'll hold on to my unlocked/overclocked phenom ii x3 for now.iwod - Wednesday, October 12, 2011 - link
Intel made most of their money with Server CPU. And If BD perform well there, then i suppose AMD still have some breathing space.Otherwise with Ivy Bridge, AMD doesn't have a single chance of surviving in the near future. Gfx are much better with Ivy, and with its video decoding engine that seems to be much better then even Nvidia or AMD.
milli - Wednesday, October 12, 2011 - link
"Sandy Bridge, which on Intel's 32nm HKMG process is only 1.16B transistors with a die size of 216mm2"But in the table below it you say it's 995M transistors.
In the AMD table, you mention '3MB' as NB Clock for the AMD Phenom II X6 1100T.