hihi ,yes and passed .Quote:
Originally Posted by GAR
Printable View
hihi ,yes and passed .Quote:
Originally Posted by GAR
Anyone played with these options which I have circled in this screenshot?
Results? effects?
http://i194.photobucket.com/albums/z...iosoptions.jpg
I was wondering the same thing.......anyone??Quote:
Originally Posted by Teccxpert
i found thisQuote:
Originally Posted by GAR
When timing problems occur between CPU and memory, your rig becomes unstable. If this happens while overclocking, you back off a bit thinking you reached your cpu/memory limits. If it happens during the initial boot of a new build you curse your bad luck and get ready to RMA your high-priced memory and/or mobo. One last adjustment you need to try before either of these actions, that might just save the day, is adjusting your clock skew. Timing issues can occur between any of the many buses in your computer, but it is a real problem when it occurs amidst the ultra high speed chatter of the cpu, memory controller and memory channels. Mobo manufactures know this and some, such as ASUS and DFI, provide settings in the bios to allow you to fine tune the memory clock and compensate for these timing issues. According to ASUS, you may be able to maintain a stable 10 to 15% performance pop by adjusting clock skew. It has also been reported that mystery compatibility problems between mobos and high quality memory modules such as Corsair and OCZ were fixed by adjusting skews.
The Bottom Line Adjustments:
Most overclockers leave clock skew adjustments at their factory default of “Auto.” The problem is that when overclocking at higher frequencies, the auto control does not adjust precisely enough to keep your overclock from crashing, hence a manual setting is often needed. Also, "Auto" cannot compensate sufficiently to resolve some memory compatibility problems.
Manual adjustments for clock skew are here for ASUS and DFI mobos:
ASUS mobos:
ASUS mobos allow you to tune each memory bank separately. Very handy, since each bank has its own transmission path and need different skews to be fully optimized.
In some ASUS mobos the bios adjustment is referred to as “DDR Clock Skew.” There is one for “Bank A” and another for “Bank B”. In other mobos they are referred to as “AI Clock Skew for Channel A” and another for “Channel B”.
Regardless of how they are labeled the choices are the same: “Auto”, “Normal” and a series of “Advance” (speed up clock - reduce setup/hold time) and “Delay” (slow down clock - increase setup/hold time) adjustments from 150ps to 900ps. These are not large adjustments - one picosecond (ps) is one trillionth of a second. Light travels only 3mm in one picosecond.
I don’t have any specific recommendations as to settings because variances between components are too wide. However, a 1 or 2 ps adjustment is not going to do much. Think in increments of 50 or 100ps starting at 100 or 200ps. It’s tough to find solid info on this adjustment (a cynical no-thanks to ASUS) but I did find this reference from Tony at bleedingedge that seems to be a good and reasonable approach. I took the liberty of including it. He was tuning his A8R32 MVP mobo. Here is what he wrote:
“For channel A I found normal or a slight delay worked best, for DDR600 delay 150ps
For channel B I found advancing worked well, 300ps for DDR600.
Now, you must tune and test with 1 dimm, tune the skew and get the ram stable as high as you can in memtest86+. Then lower the clock a little and remove the dimm, add the second dimm and test the same and tweak the skew for the second channel. Channel A is closest to the CPU and B furthest away. Do NOT alter any memory timings; only tune the skew under the config system frequency/voltage section. Once you have a base line knowing what each dimm is stable to, re enter bios and set 5 fsb lower than the lowest stable clock from the 2 dimms tested. Switch off and add the second dimm back to its slot, reboot and re enter memtest...check if it’s still stable.”
DFI mobos:
“DQS Skew Control” = “Auto”, “Increase Skew” (speed up clock - decrease setup/holdtime), “Decrease Skew” (slow down clock - increase setup/hold time).
"DQS Skew Value"= 0-255 in 1.0 increments. This is the value that is increased or decreased when you set the DQS Skew Control. Begin adjustments in the 50-255 range.
Conclusion
I was as many overclockers content to let Clock Skew ride at Auto because I did not understand how it worked. I wrote this FAQ simply because internet searches yielded a lot of references to promotional materials lauding its value - but not much as to how to actually use it. After much research I now know that it can be a vital adjustment and is an important tool in the overclocker's tool chest. I’d like to apologize to all electrical engineers if I killed any technical details in the writing of this FAQ, but I wrote this for the non-technical overclockers – like me.
i got board going lol
wasted all that time all i had to do was reset cmos manually lol
i went back to f3
f4 poor overclocker n those skew settings :shrug:
Quote:
Originally Posted by vanovich
thanks, one last question. When oc'ing do i have to go in steps or can i jump from 266 straight to 400 fsb?
Thanks
Straight to 400, tried it and it worked like a charm.Quote:
Originally Posted by kam03
nice and informative writeup there :)Quote:
Originally Posted by zsamz_
that may explain some of my recent monster clocks with Corsair PC10K sticks on Gigabyte boards
did you write this or found it on the net somewhere....if so where?
X38-DQ6, F5d beta:
http://rapidshare.com/files/60451262/x38dq6f5d.exe.html
GA-X38-DQ6Quote:
Originally Posted by Richard Dower
F5F 2007/10/04
1. Improve CPU FSB overclocking and memory performance
http://www.giga-byte.com/Support/Mot...ProductID=2665
GA-X38T-DQ6
F3G 2007/10/04
1. Improve CPU FSB overclocking and memory performance
http://www.giga-byte.com/Support/Mot...ProductID=2674
nice find, im going to install this now, be back in a bit with results.
EDIT.........
OK, so new bios did nothing for me, same max mem clock, and max cpu clock......voltage seems to be a bit higher with the same setting, had to drop it one to get the same amount of volts......all in all seems to be a bit better than F4 but preety much the same.
Out of box, put on table :D
E6850 @ GeminII
2 x 512MB Crucial Ballistix DDR2 800MHz
Gigabyte X38@F3 bios
Board clock my ram identical as P35 ( Blitz/MSI Neo-FR )
http://images11.fotosik.pl/87/19d43f11821ec5d3.jpg
Martin
ok, an issue i am running into, when i restart sometimes, the board does not boot back up, even tho it is a stable OC.......anyone else get these hangs??
One other thing, this boards voltage on the cpu jumps around, like at setting 1.5725, it jumps from 1.520-1.568 on 100 load w/ P95, on idle sit around 1.536........any way to fix this?
Heh surpise :D
OBR say max FSB with E6850 is 510MHz and lol :rofl: for me also.
510 rock stable, 511MHz no boot:shakes:
Going to try this new bios :cool: Maybe F5 can do better.
Martin
540 here no problem .http://valid.x86-secret.com/show_oc.php?id=245784Quote:
Originally Posted by DEVIL K-ce
On P35-DS4 :rolleyes: Not X38-DQ6Quote:
Originally Posted by vanovich
Martin
oh sorry my mistake ,ill take a new pic tommorrow
i just googled cpu skew lolQuote:
Originally Posted by dinos22
ok, got me a new G0 and listed my B3 on ebay, this G0 so far is priming at 3.73ghz, about 20 mins into it, with 1.47v @ load, my goal is 3.8....if i get 3.8ghz i wont upgrade my cpu for a year, that would be a long time for me lol
Any results on the new F5F Beta bios guys?
I am building a 3rd rig atm & am curious as to any performance improvements from the f5f bios...
Parts I am using are:
Q6600 G0
X38-DQ6
2 x 1gb Corsair TwinX 6400C4 (waiting on my OCZ FlexXLC 9200 to come back from RA)
Gigabyte Aurora Case (bought it cos I got the watercooling kit free)
Thermaltake 750w PSU
2 x 250GB Raid-0
Have just put PC together... planning on installing XP Pro/Vista Home Preium Dual boot...
Hopefully it performs better than my P35-DS4/q6600/6400C4/OCZ750w PC
edit:
I was lucky enough to be able to go to my old workplace & "swap" my POS Q6600 B3 for a better G0 one (B3 was not stable at 333fsb lol)
i just did its much better than f4Quote:
Originally Posted by Minnyboy
really, i didnt see much difference.Quote:
Originally Posted by zsamz_
i was told this one has all the changes i asked forQuote:
Originally Posted by kam03
will test when i get some free time
fingers crossed it's not :banana::banana::banana::banana:house like F4 :)
its much better than f4 i can clock memory 1200+ but performance is poorQuote:
Originally Posted by dinos22
i had best performance with f3 @530fsb
f5g has descent performance @ 400-480fsb but 500+ is verry poor
i cant get 10000 read no matter what i got 10200 copy with f3@ 540fsb
I experienced a loss in GPU performance(Bios F5F) on all benchmarks across the board. Setting Graphics Boost to "Turbo" and PCIEX to "1000" had no effect with F5F. Rolled back to F4 and GPU performance returned. Bios F4 clocks the memory to 1212 memory settings 266D with 2GB Corsair Dominators 5.5.5.15 2T.Quote:
Originally Posted by zsamz_
Has any one come across this problem?
http://www.youtube.com/watch?v=WVOh3CcutX8
I've had this problem with gigabyte 965P DQ6 and even after i RMA'ed the board, the replacement motherboard gave me the same problem after three weeks of usage.
Since the 3rd RMA i told them to give me a new board and ill pay the difference; thats how i got the X38 DQ6. Hope this problem isn't present on the X38 DQ6.
Yeah, its from a bad OC.....happens when my board doesnt like my ram settings.
Also on another topic, what is the best strap, A/B/C/D when using the 1/1 multi for the ram
Quote:
Originally Posted by GAR
it all depends on fsb best way to test is tryin it n test with memtest
funny thing i got 6900mbs with f3 n nothin close with f5g
yes f5f suks in performance i gonna go back to f3Quote:
Originally Posted by paulxmn
it's not a faulty boardQuote:
Originally Posted by kam03
just bad OC or some incompatibility ;)
just turn off everything and pull our the battery and short the cmos and fire it up again a couple of minutes later :)
Agree in 100%Quote:
Originally Posted by zsamz_
Martin
i didnt notice ne performance difference............
here is a screen shot with my memory running at 1/1 @ 423fsb, 423mem (4-4-4-6)
http://img455.imageshack.us/img455/6812/98257169oy1.jpg
back to f3 bios
540fsb 7 multi 1/1 5-4-4-9
everest read10016 write 9836 copy 10349 latency 50.8
in sandra i get 9124mbs
545fsb
read 10102 write 9900 copy10435 latency50.6
sandra 9202mbs
but i couldnt run 4 sticks on any of the bioses lol i didnt try anything less than 1000mhz lol
everest and sandra are rubbish benchmarks
show us something solid
like what dinos?Quote:
Originally Posted by dinos22
superpi, AM3, 3DMARK01.....Quote:
Originally Posted by GAR
they are all heavily efficiency dependent
http://img371.imageshack.us/img371/794/am3wm8.jpg
3dmark 2001 coming next.....mind you these results are with Vista 32
http://img452.imageshack.us/img452/6162/2001ob0.jpg
can you pls test these against two bioses with same settings
sure, but its got to be tommorow, its 1 am here in Los Angeles, gotta get some sleep bro.....tommorow i plan on doing lots of testing with the three bioses to see wich is best.
540fsb 7 multi i get 13s
lol GAR no one is forcing you to do it....you do it at your pace of course :D
zsamz_ that would be superfast if that was the case (flat 13s 1M at 3780Mhz)
Quote:
Originally Posted by dinos22
@ 425fsb 3.825 1/1 timings 4-4-4-12 1t same result
everest read 8424 write 7723 copy 8314 latency 57
update just did 500fsb @ 4.0 1-1 4-4-4-12 2t 12s flat
funny thing i got vcore @ 1,55v and vdim volt jumped by0,02v
and 2mb is 31s
i just tried the superpi 1.5 n i get 13.125 @ 500fsb 4.0
550fsb 7x 5-4-4-12 13.2s
everest
10137 read 10019 write 10540 copy latency 50
515fsb 8X =12.422 in spi
yesterday installed my new x38 dq6, bios f3.
read in a other forum, that after you have loaded the new bios
eg F4 of f5, you should Clear Cmos.
the procedure is apparently as follows.:
swith off the pc remove the power cord, than
clear cmos, restart ps hit del load optimized defaults.
is this the correct procedure, read somewhere else that
somebody did this and the board did not boot any more the
bios was corrupted en the back up bios also did not kick in.
seems this is not covered by the warrenty terms and conditions.
updating the Bios is at your own risk.
thanks in advance.
what is your RAM setting at Normal, Turbo or Extreme?Quote:
Originally Posted by zsamz_
here is everest with mem at 423 1/1 multi 4-4-4-6
http://img443.imageshack.us/img443/2145/15463178pu6.jpg
GAR Here Is My Everest Results:
http://i38.photobucket.com/albums/e1...ntitled-13.jpg
well, my frequency is much lower, 423mhz also i am running 4 sticks, so i guess that is the reason for the slower latency??? also im on vista maybe that makes a difference
always extremeQuote:
Originally Posted by dinos22
could someone with 4x1gb setup please run everest memory test, i want to see if my latency is higher because of 4 sticks, thanks.
GAR>>>> I was a hard core Vista user, till recently. I switched back to XP, and it just runs so much smoother, faster, and efficiently. I love the Vista theme, but it's a resource HOG, and still has too many software, and driver conflicts for me!
hi guys
i have flashed on den f5f bios and tested my new quad g0 :)
here some super pi
3825 MHZ 9x425 FSB
http://valid.x86-secret.com/show_oc.php?id=251423
http://www.hardware.feichti.com/gfx/...pi-1m-4005.jpg
4005 MHZ 9x445 FSB
http://valid.x86-secret.com/show_oc.php?id=251412
http://www.hardware.feichti.com/gfx/...pi-1m-3825.jpg
4105 MHZ 9x456 FSB
http://valid.x86-secret.com/show_oc.php?id=251419
http://www.hardware.feichti.com/gfx/...pi-1m-4105.jpg
@ 4,2 ghz i got a reboot
now i must find out what will be stable for me
Feichti
hmm..........should i go back to XP, too many DX10 games coming in november, i need vista tho.....hmmmmmmmmmmm
2 words "dual boot":eek:Quote:
Originally Posted by GAR
If F5f is slow, i have to revert back to f3 to give it another try. But here mem.performance is huuuge:
Crappy new SuperT with crappy Promos @ extreme 2.66C on 3830mhz(425*9)
2577mb/s WinRAR :eek:
http://img46.imageshack.us/img46/821...2577tb2.th.jpg
SuperPI 8m same settings 2.39min @3830мхз;)
http://img406.imageshack.us/img406/4...3825fk4.th.jpg
Everest mem bench 9730mb/52ns:
http://img340.imageshack.us/img340/4...3830sq2.th.jpg
I have the exact same proble with my gigabyte p35 DS3R mobo :down:Quote:
Originally Posted by kam03
have to take it back to the suppliers for testing AGAIN
This is normal for intel boards.Quote:
Originally Posted by Delta9b
i like speed so i gotta stick with f3 and its more stable too
f5f not stable 470-480fsb n higher fsb performance is low
Wow 64.4ns 1100 5-4-4-8 you need to change strap m8. Show us WinRAR that's where the real mem.performance can beheld, not over 10gb/s in Everest.
whats your ram running at + timmings ?Quote:
Originally Posted by zsamz_
whats the best strap??? 200/266/333/400???Quote:
Originally Posted by DStealth
You have to test it in you case and divider what's best. For me 2.66D is more stable than 2.66B and performance is also equal .Quote:
Originally Posted by GAR
cant change strap on f3 yetQuote:
Originally Posted by DStealth
to be honest my p35 gave me better performance10300 read 48 latency @ 535fsb
This better than P35 then?
So not being able to boot is normal for intel boards??Quote:
Originally Posted by GAR
my board will often get stuck continously trying to post. none of the other intel boards i have or have sold do this??
if unstable yes but some boards will cycle back to stock settingsQuote:
Originally Posted by Delta9b
gigabyte including...however i've had this happen many times when pushing too far
No, if your board doesnt like your overclock, its going to do that.Quote:
Originally Posted by Delta9b
Quote:
Originally Posted by easyrider
it should be when they fix bios
how are you guys chanching the strap? i can see them in the right menu ,any speciail key to change strap ,or does it change automatically ,when you set the ram to fx 1000?
Straps make very little difference, but some are better than others, in my expeirence. for me A gave me best performance.Quote:
Originally Posted by vanovich
New Bios is out the F5G with Improve Q6700, Q6600 CPU FSB overclocking
http://tw.giga-byte.com/Support/Moth...ProductID=2665
i will try it at the evening :)
feichti
it even happens at stock settings, definately not normal:shakes:Quote:
Originally Posted by GAR
Just put in the new bios, will do some tests later on...........
Here is my final result, not extreme but 100% stable:
450x8
Vcore BIOS 1.5V, Vcore load 1.44V
Performance Enhance Turbo
Multi 2.0A, DDR2-900, 266 strap
Timings 4-4-4-12-3-3-6-42-3-7-Auto-2T 2.1V
http://img510.imageshack.us/img510/5...3600pn2.th.jpg
3DMark06: http://service.futuremark.com/compare?3dm06=3313019
Is the team memory really that bad for this board? Have 4x1GB PC6400 4-4-4-10 and can´t go stable over 3,2 with my Q6600 G0 on the GA-X38-DQ6 even when i have only 2x1GB installed.Quote:
Originally Posted by vanovich
Was almost on the way to RMA the board when found your post about this.
Will i have better luck with OCZ PC9200 FLEX XLC? Overkill perhaps but will get a good deal so why not.
you sure it's not 200 strap manQuote:
Originally Posted by merlin76
According to BIOS:
A -> 266
B -> 333
C -> 200
D -> 400
Just got mine, after getting to finaly post it will only clock too 450FSB
RAM has done DDR1200 and past memtest fine, CPU has done 4.3GHZ
I had to flash the BIOS in windows, running F5f
So what am i doing wrong!
What hardware are you using? did you try F3?Quote:
Originally Posted by Pete
CellShock PC8000 C4Quote:
Originally Posted by GAR
X1900XTX
E6850
Tried stock shipping BIOS
did you try a different strap, straps make a big difference when scaling to a high fsb.....try a looser strap like 400 also dont clock your ram that high when trying to reach high fsb to make sure your ram is not limiting youQuote:
Originally Posted by Pete
What strap then as it gives me numbers to letters and makes no sense as there is no doucmentation online at all.
450FSB is pretty crap.
All my past P35's have stoped at that as well!
I got a cascade to go on here shortly and need max FSB
my p35 maxed out @ 575fsbQuote:
Originally Posted by Pete
my x38 maxes out @ 555 "only f3 bios"
I ordered my own X38-DQ6, cost €241!...i see F5g is now available.
ive just invested in some pc8500 ,as i need high fsb.the best way to get most out off your team mem ,is set them to auto/spd.mine wont even let me put timmings manually. so they are running 4 7 7 23 . i thought to the board was bugged until i tried other ram in.my e6850 benches 4.5 now .Quote:
Originally Posted by E.R
ok, today i started getting randon BSOD's with my board, very wierd......im thinking of exchanging this board out, its giving me alot of issues, it doesnt clock 4 sticks good at all............even at stock settings.
Quote:
Originally Posted by GAR
i'm very interested in this issue with the X38.
I'm either going X38 with DDR2 to keep the RAM I have, or get rid of the DDR2that i have along with this P5K and move on to DDR3....
I'm currently battling my P5K DLX mobo with 4 x 1GB Patriot 9600 1200Mhz. This board HATES this memory - i'm sure of it. Can't even break 1000 without constant BSOD and freezes at SPD timings 5-5-5-12
1200Mhz at 5-4-4-12 on my Commando - NO PROBLEM.
my E6850 tops out at 552MHz FSB on P35 and 547 on X38 F3 biosQuote:
Originally Posted by zsamz_
would be nice if we can get 10000+ read stable @ 510fsb like on p35Quote:
Originally Posted by dinos22
winter commin soon :) the board n cpu gonna be much cooler @ high fsb
btw new bios is only has quad tweaks or ...
same here i tried 4 sticks n not stable at allQuote:
Originally Posted by GAR
I packaged it up, a bsod while using stock settings is not worth $300 i paid for it, going to exchange it tommorow.
When at stock watch your tRD (performance level). Mine is set to 4 and that's too tight, should be 5 or 6. For high FSB (400+) 7 or 8.
For high FSB and high RAM clocks try B or D multi and set tRD to >6.
Max stable FSB for me is 460, this is with a quad and 4x1024MB. I believe CPU tops out because it doesn't matter what strap and voltages are used.
i got 485x8 running.
G0 Q6600
4x 1GB D9GMH´s
X38-DQ6
HD 2900XT
Fullcustom Watercooled
actually im running 3,2ghz@ 1,28vcore for 24/7 Setting.
But i will try max FSB and max Clock of CPU.
RAM´s made 1200-5-4-4-12@ 2T with eVGA 680i SLI
and 1066-4-3-3-8@ 1T
Now i´m testing the X38 maxxes ;-)
Using BIOS F5G, is better then F3 and have improvements to f5f
Maybe I should do a direct comparison between this board & the P35-DS4?
Both have very similar setups... both run q6600, 2 x 1gb Corsair 6400C4, 8800GTX (MSI & gigabyte), 750w PSU (OCZ & TT)... just different hdd is all...
finally got my Ultra Xtreme installed on DQ6 (was such a chore to get it installed with washers as it was like wobbly w/out them)...
Anyways, Both have same OS/Software installed on em. will run them at stock everything & then max OC (that i can with 6400C4 while I wait for FlexXLC9200 to come back from RA)...
Will report on findings with latest DQ6 bios comparo against P35-DS4
Quote:
Originally Posted by GAR
i'm gonna give it another week but gonna contact retailer tomorow i had to reinstal os 4 times in a week
when test win memtest i notice its verry slow on testing after i pass 100% i still will get bsod
gigabyte u better get a descent bois soon or u gettin this board back
i be testin the new bios out
Yesterday i recieved an X38T-DQ6 motherboard from Gigabyte.
This is their flagship DDR3 board soon to be released to the wild.
The board came shipped with F3d BIOS (if memory serves me correct).
I flashed it up to the latest beta on the website F3h.
With nothing stopping me, i set the board up straight away from extreme benching and installed everything nicely.
X38Tshot.jpg
Then i remembered about Windows install heheh :(
So there i am installing a fresh OS on LN2 for an hour, preparing the OS for benching also.
Once i got all that out of the way, i quickly moved on to the real fun part, LN2 QX6850 on X38T-DQ6
X38T185Vload.jpg
The first impressions of the board were good, Gigabyte had finally listened to the users and done away with the sill CTRL+F1 to get access to memory timings.
The BIOS was in their typical style still, but with many more voltage options (most i have NFI about). Still no additional temp or voltage readings that i could find though :(
I jumped straight up to 1.85V BIOS set, and found a pretty bad vdroop of 1.83V idle / 1.77V load!
Into Windows i went, and shot straight for the 5050 wPrime tests that i had achieved the day before on the P35 variant.
Suprisingly, even with such droop (as measured by the choke leg on underside of mobo) it ran no problems at the same speeds as previously - and with DDR3 in da house the times were rocking :)
Unoptimised 32M wPrime - new #1 QX6850 on hwbot
http://www.hwbot.org/result.do?resultId=654434
Next up i moved onto 1024M, but i unfrotunately allowed temps to creep over -130C and a froze the OS.
No problem, nothing a reboot wont fix...... usually.
The system would not reboot no matter what i tried; clear CMOS, pull out battery, power off PSU, reset while attempting to POST, etc.
All it would do is spin up and my vcore reading was 1.6V, the screen wouldnt come on.
After about 10minutes trying to coerce it back to life, im standing there poking and proding then all of a sudden *Puff* a smoke cloud and that most foul of burning smells eminates from the CPU mosfet area of the board. ****
So i pull everything appart and it's pretty obvious that the board has blown it's load:
X38Tdestruction1.jpg
X38Tdestruction2.jpg
We'll see what the next board can handle :)
I still havent tested the QX6850 to see if it survived, but im confident.
just tried f4g n same crap back to f3
10% loss in bandwidth n 57 latency instead of 50 :rofl: :rofl:
tried running 4 sticks of d9gmh n lol nice bsod
this board dont wanna run 4 sticks @ 800mhz n my p35 doin it @ 1020mhz
TM, if that was all the armaflex you used that doesn't shock nor surprise me it went like it did
This motherboard seems to be teh suxxorz, I knew I did the right choice when I canceled my order on this board and ordered maximus formula instead.
No problem here:Quote:
Originally Posted by zsamz_
http://img514.imageshack.us/img514/4...1000nm0.th.jpg
333x10
Vcore BIOS 1.35V, Vcore load 1.3V
Performance Enhance Turbo
Multi 3.0A, DDR2-1000, 266 strap
Timings 4-4-4-12-3-3-6-42-3-6-Auto-2T 2.25V
im glad i canceled my order too