Nehalem - Everything You Need to Know about Intel's New Architecture
by Anand Lal Shimpi on November 3, 2008 1:00 PM EST- Posted in
- CPUs
Cache Hierarchy
Once again, I’ve already talked about the Nehalem cache hierarchy in great detail so I’ll keep this to a quick overview.
Nehalem, like AMD’s Phenom, features a 3-level cache hierarchy. There’s a 64KB L1 cache (32KB I + 32KB D), a 256KB L2 cache (per core, unshared) and up to an 8MB L3 cache (shared among all cores).
The L1 cache is the same size as what we have in Penryn, but it’s actually slower (4 cycles vs. 3 cycles). Intel slowed down the L1 cache as it was gating clock speed, especially as the chip grew in size and complexity. Intel estimated a 2 - 3% performance hit due to the higher latency L1 cache in Nehalem.
The L2 cache also gets a hit - while in Penryn we had a 6MB L2 cache shared between two cores, Nehalem moves the L2 cache next to each individual core and reduces the size to a meager 256KB. We haven’t had a high performance Intel CPU with such a small L2 cache since the first Pentium 4. The smaller L2 is quicker, it only takes 10 cycles from load to get data out of the L2 cache.
The L2 cache acts as a buffer to the L3 cache so you don’t have all of the cores banging on the L3 cache, requiring tons of bandwidth.
The L3 cache is shared by all cores and in the initial Core i7 processors will be 8MB large, although its size will vary depending on the number of cores. Multi-threaded applications that are being worked on by all cores will enjoy the large, shared L3 cache.
Intel defended its reasoning for using an inclusive cache architecture with Nehalem, something it has always done in the past. Nehalem’s L3 cache is inclusive in that it contains all data stored in the L1 and L2 caches as well. The benefit is that if the CPU looks for data in L3 and doesn’t find it, it knows that the data doesn’t exist in any core’s L1 or L2 caches - thereby saving core snoop traffic, which not only improves performance but reduces power consumption as well.
An inclusive cache also prevents core snoop traffic from getting out of hand as you increase the number of cores, something that Nehalem has to worry about given its aspirations of extending beyond 4 cores.
Facebook
Twitter
RSS
35 Comments
View All Comments
qurious69ss - Friday, August 22, 2008 - link
You sound like one of those sad fanboys from amdzone. Tell dimentia to get a life.X1REME - Friday, August 22, 2008 - link
wow, this whole cpu is a copy of a amd cpu and you expect amd fan boys to not get amd with you, secondly this fantasy is baseless until you can compare it to an offering from the AMD team (Shanghai & Deneb). AMD is still KING with there OPTERON and most likely will be in the future with there new cpu coming soon for the server and also for the desktop.DigitalFreak - Friday, August 22, 2008 - link
Learn to spell, you goober.X1REME - Friday, August 22, 2008 - link
OK DORK, am sure you have never made a mistake (there=their) Duh. I bet your some kid all hyped up for the i7 who wishes Xmas comes early lol. Anyway it’s not a desktop chip, it’s a sever chip DUH. It’s meant to compete with the AMD Opteron chip (the best). Although Opteron will lose its crown, it won’t be to i7 but to Shanghai (AMD new latest and greatest). And like I said b4, Deneb will clear anything up out of place.The reason amd does not grab a microphone and star shouting at the top of their voice is because amd doesn't have the resources and money in comparison with Intel. If it reveals too much about its future strategy and Intel likes that strategy (like the Opteron, HT, On-board mem etc) there is a big theoretical chance that Intel could take this idea and deliver a product well before AMD. So it’s not over until amd says its.
snakeoil - Thursday, August 21, 2008 - link
nehalem fails,it was supposed to be superior to core 2, intel was against the wall this time, why?, because, the old front bus architecture was lagging more more in the server arena and becoming a bottleneck ,compared to hypertransport, so intel is forced to abandon the front side bus, but the strong point of core 2 is that because you don't have and integrated memory controller you can stuff the processor with a huge L2 cache.so, nehalem sucks in gaming,there is no way that the enthusiast is going to pay more for a processor that produce less fps that they actually have.
and the hyperthreading is a risky move, hypertrheading is known por being power hungry, and although produce gains in some applications,some servers applications actually runs slower, so in many cases the old hyperthreading had to be disabled.
nehalem is crippled for the enthusiast,and the regular user.
nuff said.
AssBall - Saturday, August 23, 2008 - link
You musta missed where Anand says several times its not intended for better gaming? It will be significantly faster than Penryn for multithreaded applications. I guess I don't see how this makes it "fail". Maybe in your fantasy world where 90% of the CPU market are "enthusiasts".snakeoil - Saturday, August 23, 2008 - link
enthusiasts drive the market you fruityassUnlimitedInternets36 - Saturday, August 23, 2008 - link
LOL this year Satan err Santa is going to take away your PC because you don't deserve to have one anymore You Jaded nerd.Gasaraki88 - Friday, August 22, 2008 - link
Thanks! I never knew there was a expert on CPU design in the house. I've learn so much from your well researched, tested and thought out comment...pool1892 - Friday, August 22, 2008 - link
first of all the enthusiast market is a very tiny niche, it would not kill intel if you were right.but you are not. the L2 of penryn (and banias) is much more like the nehalem L3 than the nehalem L2. and if you have a single threaded game it now has 8mb at similar latencies, but with a second buffer, the 256k L2, and a MUCH smaller cache miss penalty.
concerning hyperthreading: please read the article first. nehalem switches off what it does not need, powerwise. and about fiftytwo other very vaild arguments.