Announcement

Collapse
No announcement yet.

The Sandy Bridge Core i7 3960X Benchmarked Against Today's Six-Core / 12 Thread AMD/Intel CPUs

Collapse
X
 
  • Filter
  • Time
  • Show
Clear All
new posts

  • atomsymbol
    replied
    Originally posted by perpetually high View Post
    compiling the kernel with -march=haswell, -O3, 1000hz timer, preemptive kernel (low-latency), etc etc. I think it matters.
    Some notes:
    • Higher timer frequencies (1000 Hz) and a preemptible kernel can in certain scenarios result in lower IPC (instructions per clock) compared to lower timer frequencies (300 Hz, 100 Hz) and a non-preemptible kernel. Reason: It might be harder to refill the L1/L2/L3 caches in 1 millisecond than in 3.3ms or 10ms. Low-latency kernel can reduce perceived latency in for example GUI applications (such as interaction with terminals, browsers, IDEs, etc), but it may also decrease overall throughput which means that although applications feel more responsive it takes longer time to finish the tasks.
    • In cases when the number of processes running at the same time is not larger than the number of hardware threads, it most likely does not matter whether the timer frequency is 100 Hz or 1000 Hz and it does not matter whether the kernel is configured for low-latency or for throughput. Typical desktop CPUs of today can run 8-16 threads in parallel, which means that it is less likely for the kernel to run out of hardware threads when scheduling tasks compared to the case of low-thread CPUs of the past (1-4 threads). Future typical desktop CPUs will have 16-32 hardware threads, so the likelihood of running out of HW threads will be less likely than today.
    • If a machine feels unresponsive then the user should primarily take a look at how many tasks are competing for the available hardware threads and should check whether the peak number of running tasks exceeds the number of HW threads. A desktop environment indicator such as xfce4-systemload-plugin can help identify cases of high CPU utilization. The scheduling priority of build jobs (Makefiles, etc) should be kept lower than the scheduling priority of GUI applications.

    Leave a comment:


  • perpetually high
    replied
    Wow, that i7-3960X (released Q1'11) is pathetically slow for 6c/12t.

    I ran a couple tests out of curiosity on my 4c/4t Haswell i5-4670K (released Q2'13), and got a ctx-clock score of 142, which beats all the processors in this test. (3600X was 167, 2600X was 180, i7-8700K was 913, and i7-3960X was 1103)

    This is exactly why I and many others enable mitigations=off. You're leaving way too much performance on the table otherwise, especially for desktops. I think these processors are still very capable. It's also why I bother compiling the kernel with -march=haswell, -O3, 1000hz timer, preemptive kernel (low-latency), etc etc. I think it matters.

    Btw: I ran the ctx-clock benchmark with the recent Intel FSGSBASE patches from yesterday and it received the same score as without the patches. Both got a score of 142. Figured this was a more relevant test to see how the patches affected context switching, but I'll leave the benchmarking to the pros. For now, I'm leaving the patches enabled as it seems stable so far.

    EDIT: Also ran the ctx-clock on my 2010 MacBook Pro running linux and got a score of 162. This i5-520M beat out the Ryzen 3600X!
    Last edited by perpetually high; 14 September 2019, 12:01 AM.

    Leave a comment:


  • skeevy420
    replied
    Originally posted by ms178 View Post

    It depends on your tasks, if they are not using or cannot meaningful profit from AVX/AVX, then it is still fine as in many games. I've used a Xeon X5675 @ 4.2 Ghz for gaming on Windows still until this summer. As the security mitigations take more of a performance hit on these older architectures, I'd suggest to disable all of these for a gaming box.
    That. The x5687's in my current setup are just fine for most needs though it's obvious that I'm CPU limited in regards to modern gaming (2017+ games). I'll probably upgrade next year or sooner if I can find some good deals on corporate Zen workstation lots on eBay.

    Leave a comment:


  • Teggs
    replied
    I have seen comparisons that look like this before. It is what happens when someone benchmarks Bulldozer. Time has washed away the differences between the processors of that period. If 'Bulldozer sucks', then 'Sandy Bridge sucks' too.

    What this particular chart does not list is the relative prices of the 8700K and 3600X (or for that matter, the 3600). In a word: 'ouch'.

    Leave a comment:


  • ms178
    replied
    Originally posted by Zan Lynx View Post
    I guess my six core 980X is barely worth running any more. Heh.
    It depends on your tasks, if they are not using or cannot meaningful profit from AVX2/AVX, then it is still fine as in many games. I've used a Xeon X5675 @ 4.2 Ghz for gaming on Windows still until this summer. As the security mitigations take more of a performance hit on these older architectures, I'd suggest to disable all of these for a gaming box.
    Last edited by ms178; 14 September 2019, 07:31 AM.

    Leave a comment:


  • Zan Lynx
    replied
    I guess my six core 980X is barely worth running any more. Heh.

    Leave a comment:


  • Michael
    replied
    Originally posted by darkbasic View Post
    Wow, didn't realize the improvement was that big. Could you please re-test with the processors at the same clock?
    Unfortunately not. Already packed up the 3960X system again for another year and the 3600X moved on to other tests.

    Leave a comment:


  • kylew77
    replied
    So architectural improvements make a huge difference it would seem. It's not just pure core count that makes all the difference or even a little bit of clock speed difference. either.

    Leave a comment:


  • linner
    replied
    My poor 2680's... it's all I can afford

    Leave a comment:


  • darkbasic
    replied
    Wow, didn't realize the improvement was that big. Could you please re-test with the processors at the same clock?

    Leave a comment:

Working...
X