July 22, 201114 yr Sound promissing....the quad chanels is helping a lot.http://www.techpowerup.com/149464/Core-i7-3960X-About-47-Faster-On-Average-Than-Core-i7-990X-Intel.html
July 22, 201114 yr Sound promissing....the quad chanels is helping a lot.http://www.techpower...990X-Intel.html Sounds indeed promising! Can I put this CPU in my current system or do I have to get everthing new? Best regards, Steffen Fight time: NGX 737-700: 37,0h; -800: 47,2h
July 22, 201114 yr You are funny, Steffen!Why not start with overclocking the i5 you have humming around at 3.7 Ghz?
July 22, 201114 yr Look at the pics attached. The difference is significant only for AVX ready benchmarks and memory benchmarks like SysSoft Sandra that like memory channels. In Cynebench etc, it's only a 15% faster (from the Sandy Bridge IPC)At the risk of being a pest I'll say it again. This is a massive waste of money for FSX
July 22, 201114 yr Author Dario, I'm with you on FSX but some of us play other games also.Soon we will see 3.0 GPU on the market as we can already see mobo with PCI-e 3.0, how long before we see quad chanels memory...we need to see how far (GHz) some will be able to push this new CPU, will it overclocked better compare to the 2600K?Anything new will help FSX as the Sandy Bridge did compare to the X58 platform.
July 22, 201114 yr Dario, I'm with you on FSX but some of us play other games also.Soon we will see 3.0 GPU on the market as we can already see mobo with PCI-e 3.0, how long before we see quad chanels memory...we need to see how far (GHz) some will be able to push this new CPU, will it overclocked better compare to the 2600K?Anything new will help FSX as the Sandy Bridge did compare to the X58 platform.Not anything Alain. IPC, overclock would, but there's no reason to believe any of those will be any better. It's the same architecture.
July 22, 201114 yr It's the same architecture.Bingo ___________________________________________________________________________________ Zachary Waddell -- Caravan Driver -- Facebook: http://www.facebook.com/zwaddell Avsim ToS Avsim Screenshot Rules
July 22, 201114 yr TPU has really misrepresented the facts first presented by Donanim Haber in this piece. As you can see by clicking on the 2nd of 3 slides, SB-E is only 12-15% faster than Westmere at the same clockspeed, in compute-intensive tasks, and much faster when you throw in AVX or stress memory bandwidth. FSX is neither memory bandwidth bound on modern CPUs nor does it make use of AVX so SB-E will be nowhere near 47% faster than current 6-core Intel chips. As for Flight, that remains to be seen but I for one would be shocked if it made use of AVX or were terribly constrained by memory performance on modern platforms.
July 22, 201114 yr Not anything Alain. IPC, overclock would, but there's no reason to believe any of those will be any better. It's the same architecture.Qualitatively yes, but quantitatively no: Even if the 6 CPUs and 4 RAM channels don't give additional performance, the 3960X almost doubles the L3 cache WRT the 2600K from 8 to 15 MBy. Note the increase in CPUs from 4 to 6 would only "justify" a 12 MBy L3 cache, so Intel is throwing in an extra 3 MBy (.5 MBy per CPU) of L3. The larger L3 will increase the cache hit ratio, thus giving the appearance of faster RAM access.Cheers,- jahman.
July 23, 201114 yr Qualitatively yes, but quantitatively no: Even if the 6 CPUs and 4 RAM channels don't give additional performance, the 3960X almost doubles the L3 cache WRT the 2600K from 8 to 15 MBy. Note the increase in CPUs from 4 to 6 would only "justify" a 12 MBy L3 cache, so Intel is throwing in an extra 3 MBy (.5 MBy per CPU) of L3. The larger L3 will increase the cache hit ratio, thus giving the appearance of faster RAM access.Cheers,- jahman.L3 cache doesn't seem to be helping much in those benchmarks does it? Same 2600K vs 2500K debate
July 23, 201114 yr L3 cache doesn't seem to be helping much in those benchmarks does it? Same 2600K vs 2500K debateIntel keeps increasing L3 (an apparent repeat of the 2600K vs 2500K debate) instead of keeping L3 at 1.5 MBy/CPU as in the 2500K, thus shrinking the die size, increasing wafer yield and thus profits. So L3 must be worth its gates for certain types of apps.Cheers,- jahman.
July 23, 201114 yr Intel keeps increasing L3 (an apparent repeat of the 2600K vs 2500K debate) instead of keeping L3 at 1.5 MBy/CPU as in the 2500K, thus shrinking the die size, increasing wafer yield and thus profits. So L3 must be worth its gates for certain types of apps.Cheers,- jahman.I don't doubt that's the case. Not for gaming in general though
July 23, 201114 yr You are funny, Steffen!Why not start with overclocking the i5 you have humming around at 3.7 Ghz? Cause I don´t need it at the moment. if I fly FSX, it´s on 4.6. Best regards, Steffen Fight time: NGX 737-700: 37,0h; -800: 47,2h
July 23, 201114 yr Large L3 caches have been demonstrated time and again to be worthwhile in server and HPC workloads, with consumer application benefit at virtually zero.
July 23, 201114 yr Interesting, yet wrong socket then? Or is it impractical to design two different chips with different L3 cache sizes for user vs. server markets? (I thought the Xeon chips were for servers.)Cheers,- jahman.
Create an account or sign in to comment