Official HWBOT Partner

Submission Details

10 min 31 sec 657 ms with
Phenom II X4 965 BE @ 6748.70 MHz

Images

Hardware Details

Processor

Model
Phenom II X4 965 BE Deneb (4 Cores Active)
Cooling
Liquid Nitrogen
Cores
6,748.70 MHz  (98%)

Memory

Product
Project X
Cooling
Air (Stock)
Type
4 GB  DDR3 SDRAM Speed @ 915 MHz
Timings
tCAS: 6.0  tRCD: 6  tRP: 5  tRAS: 17

Videocard

Model
Radeon HD 5770 (Juniper) (TeraScale 2 (Evergreen))
Cooling
Air (Stock)
Speed
850.00 MHz  /  1,200.00 MHz

Motherboard

Cooling
Air (Stock)
Chipset
990FX

Comments

BeepBeep2 commented on own score:
Kind of angry, Board is extremely slow. 4-5 runs failed at or after loop 19, one failed at 23, was on track for 10:29.xxx. CPU is good for 6800 32M unless I degraded it too much.
BeepBeep2
BeepBeep2

Max CPU-NB possible, I tried to run 6.78/4.6/915 6-6-6-18 with CWL = 6 (2sec faster, but needs much higher vdimm) but this run was with CWL = 7.

 

Infra, the runs after this had me on my toes because I had you beat, but the rig didn't want to have it. -_-

I.nfraR.ed
I.nfraR.ed

Wow, so close. The cpu is great, but c5f is inefficient board, you're loosing around 7-8sec I think. Maybe more.

knopflerbruce
knopflerbruce

Lmao. I had one like this with K8 a couple of years ago. In the lead even after 24th loop finished, but lost by 0.1s in the end... to massman I think it was :P

 

You're still more lucky than me with C5F - my board has cold boot bug at -90, and seems to coldbug at like -180c.

 

No gain my lowering TRC? I can run 11... as tight as it goes.

I.nfraR.ed
I.nfraR.ed

Look at my best Deneb runs. It's not because I can't run trc=11 :P

 

Oh..and btw I still don't know how to master the final loop. Usually lose some time there when I'm far ahead after the 24th loop.

 

Beep, if you compare our scores, you're only faster at the first 2 loops, due to pure speed.

But..my final loop is bad.

 

Looking at Calathea's score, I'm 1 sec faster than him until the last loop, but at the end our times are very close. That's why I was disappointed, because I had a proper run, but lost it - 10:29.xxx.

BeepBeep2
BeepBeep2

Yeah, that is interesting. I was 1 second faster than you in the last loop. As far as TRC, I always thought it was most efficient to run CL + TRAS, chew* told me that TRC = 11 was slower but could allow higher clocks with some DDR3.

 

I'm not sure what is up with TRC = 15, must be faster? Not sure.

 

I know on K8 + NF590, I found that TRC = between 13-15 was fastest no matter what primary timings you used.

I.nfraR.ed
I.nfraR.ed

Yes. Don't remember if I've tested all values between 11 and 24, but probably most of them and 15 was fastest for me. Since then, I always run trc=15.

Even at 940 with the gtx2, doesn't hurt my clocks.

BeepBeep2
BeepBeep2

What mattered to me the most was CAS Write latency though, default is 8 I think but 6/7 are just so much faster, can improve time by 3-4 seconds even.

When I get back from vacation in mid July I'll have more time to test, but having only 30L on hand at a time is a little inconvenient, goes fast when things aren't going as planned.