Spi_32m break 10min with ONLY 5Ghz, fastest 32m perclk ^_^

[VictorWang]

X6800 box retail L627A872 @ -155c loaded, vcore=1.86v
Asus P5B-Deluxe wifi/ap, 1.04G, bios=1004, vmch=1.86v
Team Xtreem DDR2-800 C3 EPP 1G*2, vdimm=2.8v
RaptorX/5700 PCI/PC P&C 1kw.

----------------------------------------------------------

thx to my great new Alu_pot, my x68 max suicide clk=5.2G, but 5.1G pi_32m benchable

[K404]

Told you you`d manage it! I knew there`d be 625MHz on the RAM

Congrats!!

[Moonman]

Congrats
Sick speed.

[VictorWang]

Told you you`d manage it! I knew there`d be 625MHz on the RAM

Congrats!!

today is just for brief LN2 testing of my new x68 (1.65v 4G sp2004 on Air ).
'coz my LN2 has only 6L left
will fill up my 50L dewar in this week.

I've tried 393*13=5109mhz, ram=1179 4445, but only 1sec faster than my 5Ghz, 9min55s, so I decided not to post that

higher ram_clk is always preffered for 32m running as I told long time ago.
even 500mhz 3-3-3-8 is slower than 1200 4445

[LittleDevil]

You have really nice CPU there, nice results

regards

[eva2000]

wow Vic I got same batch cpu too

[VictorWang]

wow Vic I got same batch cpu too

congratz my line#=1986
hope can hit 5.4G on 975x, 965x is really hard for 5ghz+ overclking

[eva2000]

Thanks

which line is that # ? 2nd line ?

mine is #1746 - bought cpu new locally and got their last one in stock .... runs extremely cool at 3808mhz at 1.55v prime at 59-62C load and at 4092mhz at 1.65v idles at 43-45C and load no more than 62-65C!

why hard for 965 ? due to strap change at 400fsb for decent times ?

[VictorWang]

Thanks

which line is that # ? 2nd line ?

mine is #1746 - bought cpu new locally and got their last one in stock .... runs extremely cool at 3808mhz at 1.55v prime at 59-62C load and at 4092mhz at 1.65v idles at 43-45C and load no more than 62-65C!

why hard for 965 ? due to strap change at 400fsb for decent times ?

yup, mine 1.65v sp2004, watercooling, ambient=16c.
coretemp=66~68c, on Asus ROG 680i

don't know why, but... 4.1G chip on 975x can only run about 4G on 965.
especially e67 , x68 and qx67.
maybe asus' 965 mobo needs GTL mod like master The Stilt done

[NiCKE^]

Oh looks so good! Now I want to play with LN2 bad

[eva2000]

yup, mine 1.65v sp2004, watercooling, ambient=16c.
coretemp=66~68c, on Asus ROG 680i

don't know why, but... 4.1G chip on 975x can only run about 4G on 965.
especially e67 , x68 and qx67.
maybe asus' 965 mobo needs GTL mod like master The Stilt done

yeah my ambients between 24-34C lol

with stock p5b dx max was 4069mhz at 1.62v idle/1.59v load until i did pencil vdroop mod for 4092mhz at 1.65v idle/load

Same as you all my E6600/E6700 B1 did 40-90mhz better on 975x than Asus P5B DX

[before]

Oh god! Victor, you've just taken my own aim... after getting the lowest clock ever to break 10s barrier @1M, I wished to achieve the lowest clock ever to break 10mins barrier @32M

Congrats! With such a hell of memory kit, you'll own this WR for ages... no doubt about that!

[wa77ss]

625MHz 32MB is very nice at 4-4-4-5!

I hope you got a fan on those RAMs though!

2.8v =

[blind_ripper]

great pi score vic , nice rams 2
hhhmmm btw switch from a 965chipset to a 680i will this gain some mhz 2 ?
hope so .
hopefull ul get great clocks with LN² on ure rest off cpu's

[hey]

nice performance/clock..

[NickS]

Awesome result, especially for 965P

[HermS]

Great benching as always, big clocks are nice but to bench those speeds with efficiency is all the more impressive.


How do we calculate the clock for clock efficiency?
Is it (MHz) x (seconds) and the lower the result the better?

eg:- (taken from hwbot top 5)

1. OPB - 5504 x 578.810 = 3185770.24
2. oc_windforce - 5455 x 580.17 = 3164827.3
3. VictorWang - 5000 x 596.810 = 2984050
4. hipro5 - 5328 x 599.56 = 3194455.6
5. Memoryextreme Team - 5240 x 612.14 = 3207613.6

[massman]

How do we calculate the clock for clock efficiency?
Is it (MHz) x (seconds) and the lower the result the better?

eg:- (taken from hwbot top 5)

1. OPB - 5504 x 578.810 = 3185770.24
2. oc_windforce - 5455 x 580.17 = 3164827.3
3. VictorWang - 5000 x 596.810 = 2984050
4. hipro5 - 5328 x 599.56 = 3194455.6
5. Memoryextreme Team - 5240 x 612.14 = 3207613.6

Shouldn't that be (seconds) / (MHz) ??

[HermS]

I don't know, that's why I asked hehe.

I know they were using (MHz) x (seconds) to see efficiency in 1m so would it be the same for 32m I don't see why not... but I'm no mathematician/statistician.

[Gautam]

Shouldn't that be (seconds) / (MHz) ??

No, MHz*seconds. Pi times and clock speeds do not follow a linear relation, rather an inverse one.

[filmbot]

Victor, are those the Geil sticks you showed off a little bit ago?

[dinos22]

good God Victor you are on fire

who said 2:3 ratio was slow

[VictorWang]

thx guys!

for this 5Ghz result,
I've done lots of Warming-up excersices

my Geil can run at lower volt. @ 1200mhz but... not good when > 1233mhz even if I flash Team's SPD to Geil...
so...maybe "Good Ram" needs to be re-defined : higher clk with good timing is preffered but not "lower volt".
I think it's same with "Good CPU", good on air not means good under cold.

here are some more reults for ur refference:
1. 3.75Ghz=12min30


2. 4.95Ghz=10min13sec


3. 5.1Ghz=9min55sec



so u guys can figure out how important RAM clk is to achieve such efficiency.
all timings I set in bios, no more tweak and no optimized OS.
maybe one day, break 9min50s with 5Ghz is prossible

[Gautam]

2:3 on 965 must be the fastest combo possible (except 2:3 on 975 if anyone can swing it at such clocks )

Great times Victor, and you're absolutely correct in saying that good air CPU's aren't always good subzero CPU's, and memory that does well at low voltage doesn't always do so well at higher ones.

[dinos22]

2:3 on 965 must be the fastest combo possible (except 2:3 on 975 if anyone can swing it at such clocks )

Great times Victor, and you're absolutely correct in saying that good air CPU's aren't always good subzero CPU's, and memory that does well at low voltage doesn't always do so well at higher ones.

heheh funny you say that because people were claiming without proof of course that 4:5 ratio is the fastest


clearly Anus needs to address the clocking limitation on 4:5 ratio to match the clocks on other ratios for high FSB benchmarking