- Jan 20, 2008
- 522
- 7
- 81
This is obviously about single thread vs multi-thread performance.
Option 1 is to run single x5650 oc'ed to 4.5 range, I'm already at 4.1Ghz with plenty of voltage headroom to spare. I've seen some folks do 4.7 on air, so I plan to push the OC as much as I can. Will aim for aprox cinebench r15 multithread score of ~1000+/-
Option 2 is to grab a $125 dual LGA1366 mobo used on ebay and run 2 x5650 at stock frequency 2.6Ghz, they turbo up to 3Ghz which should help some single thread performance workflows. This should yield aprox cinebench r15 multithread score of ~1385+/-
Also adding an EVO 950 PRO via PCIe adapter to the mix to eliminate dick io bottlenecks (not using as a boot drive). I am mostly targeting development type workloads, so Visual Studio / Java / SQL / Gradle builds etc.
Very hard to say whether reduced single thread performance will end up hurting me if I go with option 2 or if I'll be better off in most situations with a much, much higher clocked single CPU.
I think its impossible to know for sure unless running some real world benchmarks comparing the 2 options
But generally speaking (or in most RL workflows) would option 1 or option 2 be faster?
Option 1 is to run single x5650 oc'ed to 4.5 range, I'm already at 4.1Ghz with plenty of voltage headroom to spare. I've seen some folks do 4.7 on air, so I plan to push the OC as much as I can. Will aim for aprox cinebench r15 multithread score of ~1000+/-
Option 2 is to grab a $125 dual LGA1366 mobo used on ebay and run 2 x5650 at stock frequency 2.6Ghz, they turbo up to 3Ghz which should help some single thread performance workflows. This should yield aprox cinebench r15 multithread score of ~1385+/-
Also adding an EVO 950 PRO via PCIe adapter to the mix to eliminate dick io bottlenecks (not using as a boot drive). I am mostly targeting development type workloads, so Visual Studio / Java / SQL / Gradle builds etc.
Very hard to say whether reduced single thread performance will end up hurting me if I go with option 2 or if I'll be better off in most situations with a much, much higher clocked single CPU.
I think its impossible to know for sure unless running some real world benchmarks comparing the 2 options
But generally speaking (or in most RL workflows) would option 1 or option 2 be faster?
Last edited: