HT on vs HT off on the RIVE: Think again?

[HalloweenWeed]

I just did some testing W/the following settings:
i7-3930K
Turbo 46
Vcore +0.02 (variable)
Speedstep, C1E, C3, C6, & C7 enabled
VCCSA 1.25V
VTT 1.15V
LLC medium
CPU current capability: 130%
Bclck & strap both 100
4.6GHz


And I found that W/HT off:
LinX PS 40K passed 3x 195Gflops (219sec) temps 79-74-70-66-70-74 peak.
CPU case temp 67 max.
MOSFET area temp hit 80C hotspot.

HT on:

LinX took about 50% longer (388sec), 4x 110Gflops, temps 75-70-68-65-67-71 peak.
CPU case temp 63max.
MOSFET area temp hit 76C hotspot.
No downclocking occurred.


Conclusions: All temps are about 3*C cooler, LinX runs at 56.4% Gflops with HT on. FYI.

Personally, I don't think the per-thread performance hit is worth the extra threads. Also, I think some part of the departure of some other ppls opinions on my OC was they were comparing it to raw MHz W/HT on. I suggest we all try to be aware of this for future debates.


I would also be interested in ppls posting some benchmarks of all types showing the diff W/HT on vs off, and especially those that rely more on multiple threads, like encoding. Please?

[LiveOrDie]

Heres mine my TEMPS suck but like they say u never run full load so your temps will never get that high in normal use, Download CINEBENCH you will see with H/T on you will get better performance.

[HalloweenWeed]

Yeah but how much better? Cause if it isn't real significantly better, the performance gain on single-threaded apps and games will trump that. Oh well, I guess I should see for myself, I will download Cinebench, thanx.

[mdzcpa]

I still prefer HT. Then again, I work on my rig every day with multi threaded apps. That's why I sprang for X79 and SB-E (and why I used to run X58 & 990x). This platform is a crazy mutha workhorse. For just gaming, to be honest, a simple quad core SB would have done it for me. But I love this system now. I'm playing around with Ram cache and Ram disk now too.