Details of New Intel Dunnington and Nehalem Architectures Leaked 147
Daily Tech is reporting that details about Intel's new processor models were leaked over the weekend. Both the six core Dunnington and Nehalem architectures were featured in this leak. "Dunnington includes 16MB of L3 cache shared by all six processors. Each pair of cores can also access 3MB of local L2 cache. The end result is a design very similar to the AMD Barcelona quad-core processor; however, each Barcelona core contains 512KB L2 cache, whereas Dunnington cores share L2 cache in pairs. [...] Nehalem is everything Penryn is -- 45nm, SSE4, quad-core -- and then some. For starters, Intel will abandon the front-side bus model in favor of QuickPath Interconnect; a serial bus similar to HyperTransport."
Re:6 cores times 3MB = 16MB? (Score:5, Informative)
Note: if you're tempted to mod this up, don't. I rehashed the summary.
Re:6 cores times 3MB = 16MB? (Score:3, Informative)
Re:6 cores times 3MB = 16MB? (Score:4, Informative)
It seems that 16 MB of L3 cache is shared among all 6 processors. Then, each pair of cores has 3 MB between them.
So, 16MB L3 + 3 (pairs of 2 cores) * 3MB L2 = 25 MB total cache.
Re:6 cores times 3MB = 16MB? (Score:3, Informative)
that means we have 9 MB of L2 cache (total) and an additional 16 MB of L3 cache.
now i need to RTFA
Re:Dunnington and Nehalem? (Score:5, Informative)
Re:QuickPath vs HyperTransport (Score:4, Informative)
Re:Wow (Score:5, Informative)
In order to spike both cores, I need to start something like a compiler or video encoder, which is going to also eat I/O time. Its the I/O that slows down WoW more then the CPU usage. Since adding four more cores drastically increases my parallel processing power (which I don't need more of now), and doesn't do a thing for my I/O throughput (which I do need more of), its not really all that helpful.
Thats why this doesn't excite me a whole lot. We were already at a spot where a single core is more then fast enough for a majority of mainstream users, and now we're going to give out six of them? Other then being able to run spyware more effeciently, whats actually being gained?
(There are people who will benefit from this type of thing, of course. I just don't see the mainstream market as part of that group.)
Re:Intel still playing the Chuck Norris of vendors (Score:4, Informative)
Re:Wow (Score:3, Informative)
In server land, the more cores you jam on a CPU, the fewer blades you need on the rack. The fewer blades on the rack, the greater the TPS on that rack, the more efficient the server farm.
WoW won't use all the cores, but Yahoo!, Ebay and Google definitely will.
Re:Wow (Score:4, Informative)
Not sure about Intel, but in AMD's case, it was cost recovery for quad core chips where one core had a defect. They just zap that one so it doesn't show up and sell a perfectly good 3 core chip.
Re:QuickPath vs HyperTransport (Score:4, Informative)
One of the most impressive things about Quickpath is its self-calibration circuit. Makes making PCB's a lot easier and variations easier to deal with.
Re:Intel still playing the Chuck Norris of vendors (Score:3, Informative)
It doesn't help that in most benchmarks, AMD has been trounced by Intel this past year.
http://www23.tomshardware.com/cpu_2007.html [tomshardware.com]