SandyBridge 2D madness - let's tweak for the best 1m, 32m, pifast and wprime!

[Eldonko]

Hard to say if it is UD7 TBH, if you look thru the top runs at hwbot the main differences in times come from memory. If you have an 1100mhz 7-7-6 kit of hyper you can get those 4-5 seconds over those running bbse or similar at 7-9-7 or 7-10-7 regardless of board. For example Brian's run is on par with those with over 100mhz extra cpu speed due to the memory. Only way to tell for sure is a few more direct comparos like the one you did.

http://hwbot.org/hardware/processor/...=cpubenchmarks

[Pt1t]

Makes sense to me Jody. Gives me an excuse why Im not at 5m 56s at least.. :P

And me on m4e , 5.6 is OK with PSC to 1100 ... but no way to finish a 32M with hyper at 1100 7-7-7- . And at 5.5 with that ram clock , 24 loop are OK

[dinos22]

what about these two runs, no one is even close yet?? cheapseats is on UD7 i think


i have to check with the P8P67 DLX i have here again but last time i did i couldnt see any difference, not sure if they are same or not


can you also try compare auto subs with only main timings set and TRFC and command rate, maybe that ratio sucks on that bios no idea, try 21.33 with higher ram clock so you can compare to some of the runs in this thread iwth uD7

hi guys, here's a couple of runs for 5ghz and my max run so far @ 53x102.5
i started to get problems in 32m with my OCZ > 1093mhz 7-7-6. up to that is ok
think i'll need to use 54x multi to get any closer to 6mins

estimated for 7-7-6 + subtiming:

53x103.7 5495mhz 1110mhz @ 5m59x
54x102.0 5508mhz 1088mhz @ 5m59x

needed 1.53+ vcore @ 5430mhz, so lowest cpu clock is better

2600k L042B074 + H70
UD5 / bios F6d
XPsp2

all manual subtiming, same for each run


100.0x50 (100.2 bios set)




102.1x49 (102.3 bios set)



[/IMG]

[3oh6]

what about these two runs, no one is even close yet?? cheapseats is on UD7 i think


i have to check with the P8P67 DLX i have here again but last time i did i couldnt see any difference, not sure if they are same or not


can you also try compare auto subs with only main timings set and TRFC and command rate, maybe that ratio sucks on that bios no idea, try 21.33 with higher ram clock so you can compare to some of the runs in this thread iwth uD7

cheapseats is on UD5, says right in the post you quotes

i just ran 2133 at 7-7-6 5.4GHz, and it is 4 seconds slower than the M4E...higher mem clocks are showing a bigger gap. with subs all AUTO, it is even bigger (2 more seconds) as the M4E runs about as tight as you can already. i have to tighten everything up manually on the UD7.

[Splave]

sorry to break the post flow but you guys with 5.5ghz+ chips, was it relatively easy to get it there? IE is it pointless to try to tweak extra multis out chips that struggle to post at x51?

[3oh6]

sorry to break the post flow but you guys with 5.5ghz+ chips, was it relatively easy to get it there? IE is it pointless to try to tweak extra multis out chips that struggle to post at x51?

my chip is 100% stable at 5522MHz (1.50v vCORE) on water or phase, but wont validate at 5540MHz.

i can run 55x100.4 all day with all cores+HT, as well as 108.3*51, and any combination in-between. on both the M4E and UD7 (haven't gotten to 108851 yet), this is rather easy.

not sure if that is what you are looking for or not, but with this chip, there is no such thing as a finicky multi...it just has a brick CPU frequency wall.

[Eldonko]

Ha, different finding for me jody. For me up to 54x was easy, all it took was vcore. Above that took a bit of trial and error checking what bclk was going to work with what multi and so on. Every multi doesnt just work or not work, you have to find the winning combo.

[SandyTunnel]

Beats my 5.27 GHz Gulftown

[chew*]

am I going crazy here with the UD7, or is it 3 seconds slow compared to the M4E? anyone else with both boards seeing this discrepancy?

same OS, same Aida bandwidth, same everything, tightened down memory timings as best i can on the UD7, and it is 3 seconds slower...

100*54 // 933 7-7-6 // same xp sp3 OS/copy wazza/etc...

ASUS M4E - 6m 14.609
Jiga P67A-UD7 - 6m 17.579

ASUS M4E
[img]http6-4.png[/img]

then there is Brian's 5.6GHz run at 5m 56.xxx
http://www.xtremesystems.org/forums/...&postcount=143

VS

GHZ's 5.6GHz run at 5m 59.xxx
http://www.xtremesystems.org/forums/...&postcount=168

again, different OS, and not the same person so nothing can be exact, but the same 3 seconds between the M4E and UD7 show up. coincidence?

I had deduced i was losing 3 secs due to lack of the cache you guys have with 2600K versus the 2500K i have.

Could be right though......I thought it was like 1 sec faster than my pro though and my ud4 is hurt so can't test that.

[Alex-Ro]

I am vey curious about that aspect too,maybe someone with both processor can make a comparation between 2500K and 2600K same settings,same OS...

[chew*]

I am vey curious about that aspect too,maybe someone with both processor can make a comparation between 2500K and 2600K same settings,same OS...

i'm gonna test it when i get my hands on one.

[Alex-Ro]

Good luck catching a strong CPU.Latest run with 2500K,subtimings on auto,main latencies very relaxed,had some issues with asrock mobo,maybe tightening them up and a good wazza i will reach sub 6M

[Sniper]

My 2600K is easy with 54, need quite alot of volts with 55, and can only run 56 without ht. 5640 is max I'v gotten in sp1m on water. Havent tried very hard for 57, but doesnt seem likely as I'm already struggling with 56.

[dinos22]

cheapseats is on UD5, says right in the post you quotes

i just ran 2133 at 7-7-6 5.4GHz, and it is 4 seconds slower than the M4E...higher mem clocks are showing a bigger gap. with subs all AUTO, it is even bigger (2 more seconds) as the M4E runs about as tight as you can already. i have to tighten everything up manually on the UD7.

sorry im blind lol

i'll test both boards in the next few days see how it goes

[dinos22]

can you do some tests at lower clocks jody say 5Ghz with 18,66 and 21,33 multi if oissible on your UD7

[chew*]

Here's my take on effeciency with some of the higher latency stuff on P8P67 PRO.

Chip is bclk challenged ( like it hates even 100 at max clocks ) so did what I could considering.

The waza's not very strong, room for improvement.

[Dumo]

Sub 5.56ms all day with 58X

[chew*]

Jody, might this be your problem

http://www.youtube.com/watch?v=bQIqUL5JnW0

[3oh6]

Jody, might this be your problem

http://www.youtube.com/watch?v=bQIqUL5JnW0

i'm not sure exactly what the video is showing except that it shouldn't be able to run those timings...
but by tightening or loosening those timings, i can increase and decrease performance in 32M...

7   7
7   7
6   6
20  20

20  20
1   Auto
1   Auto
1   Auto
20  20
6   Auto
74  74
1   Auto
20  20
1   Auto

2   2
33  33

with auto:


with manually set tight timings:


HWInfo is also picking up the tighter timings. So if the timings are dummy timings, I shouldn't be getting
increased performance. With those set to auto, performance is even worse.

*and the closest comparison i have on M4E showing the UD7 about 3 seconds slow again
the slightly higher CPU clocks are offset by the higher memory clocks of the UD7'ish:

[dinos22]

still investigating why but tightening subs mildly even i start to get slower, tFAW and tWL seem to be triggering something from early findings

hoping you can do 5GHz runs

[3oh6]

still investigating why but tightening subs mildly even i start to get slower, tFAW and tWL seem to be triggering something from early findings

hoping you can do 5GHz runs

i can do 5GHz runs, but i don't have them done on the M4E...you want straight up 100*50, 1866 6-6-5 and 2133 7-7-6?

[dinos22]

yeah

[drnip]

Best I have done. My RAM blows balls. Using a Asus Pro Micro and the 2133MHz RAM I do have isn't compatible with it. Using some cas 6 @ 1600 RipjawsX on this run. Was thinking of ordering the ATX version of this mobo but going to wait until April to see how this Intel chipset dilemma pans out.

6m4s.jpg

[3oh6]

here you go boss...make it fast

100.2 set in BIOS.

[dinos22]

weird dude

i am in 6m33s and not on my pi OS yet