Official Gigabyte X48T-DQ6 Info Thread

[Supershanks]

Nice to see you up again

noticed that the replacement board LED lights are working normally.

you mean the pwm leds?? Thinl I mentioned rthe other week mine were going out just as the verified dmi message came up.

i will post more tomorrow evening since i need to do some house cleaning for a house that we are selling in a few days.

Aye well you will buy these expensive CPUs

luck

[Ace-a-Rue]

back...gave up on cleaning the house...i just couldn't stay away!

i am impress with the board's performance....definitely DDR3 is a PLUS, at least on this board.

i did some benching at a moderate overclock on this new 9650...i used 1T and 2T...there is a small difference between using the two...the ultimate difference was, i could run wPrime but i could not complete 3D2001...not until i change to 2T would it complete a successful run.

the difference between 1T and 2T at least for read speed is 2%...not a whole lot when you consider that i can complete 3D2001 run by just lowering the command rate to 2T.

also, notice that i was able to attain a slightly better 3D2001 at a LOWER Mhz on the QX9650...1st 3D2001 slide is the Q6600@3.9 Ghz (memory speed 868 versus 736 Mhz in the 2nd slide) while the 2nd slide is the QX9650@3.68 Ghz.

a explanation for the better performance at a lower Mhz is due to being able to use CAS 6 and PL=7...

[Supershanks]

Nice to see you back again Ace ,

back...gave up on cleaning the house...i just couldn't stay away!

Understand

i am impress with the board's performance...

Yes me to , though I think it's a happy meld of the board , ddr3 & 45nm with high FSB capability which gives a lot more flexibility in memory clocking. It's a good while since i've had this much fun

Yoiur Read Speeds look a little low ?? I can't see immediately why compatimg memsets.

That's something i'm finding quiet odd with everest. On the Asus P5E3 & on initial runs with earlier Everest versions 9x400 @1600 6-6-6-18
Compare this with a simlar recent clock with the latest Everest,


.............Asus.......Gigabyte
Read....10,639........10,317
Write.....7,221........10,126
copy......8,200.........9,940
Latency..47.8ns,,,,,,,51.7ns

Now It might be that I'm getting to grips with the timings a bit more, but hell I ain't that good Something has happened with the write Bench.

[Ace-a-Rue]

Yes me to , though I think it's a happy meld of the board , ddr3 & 45nm with high FSB capability which gives a lot more flexibility in memory clocking. It's a good while since i've had this much fun

Yoiur Read Speeds look a little low ?? I can't see immediately why compatimg memsets.

i thought a 10.4K read speed was decent at 1T...what reference point are you using to say that?

That's something i'm finding quiet odd with everest. On the Asus P5E3 & on initial runs with earlier Everest versions 9x400 @1600 6-6-6-18
Compare this with a simlar recent clock with the latest Everest,


.............Asus.......Gigabyte
Read....10,639........10,317
Write.....7,221........10,126
copy......8,200.........9,940
Latency..47.8ns,,,,,,,51.7ns

i presume the difference is the bios and maybe a better binned chipset...what do you think?

Now It might be that I'm getting to grips with the timings a bit more, but hell I ain't that good Something has happened with the write Bench.

as you get more boards under your belt or "notched" in your experience belt, generally goes a more tweaked performance...it means that your are teachable.

[Supershanks]

was just looking at the read/wraite benches as a ratio , for example
My Bench Read 10317 Write 10126 Ratio 98.15% Your 10405/9791 = 94.10% Aboiut 400 points down.

I'm not sure i'm still a bit sceptical that the Everest write bench hasn't shifted, a bit like the old sandra timings slipped & changed through different versions.

.it means that your are teachable.

Wait till I tell the wife that

[Ace-a-Rue]

was just looking at the read/wraite benches as a ratio , for example
My Bench Read 10317 Write 10126 Ratio 98.15% Your 10405/9791 = 94.10% Aboiut 400 points down.

I'm not sure i'm still a bit sceptical that the Everest write bench hasn't shifted, a bit like the old sandra timings slipped & changed through different versions.

i am at the top end for my QX9650 in FSB...looks as if 480 is toooo much (can boot into windows but can't complete 3D2001)...trying a run at 475 FSB.

now, i did have the write speed come up since then as i climbed the FSB ladder to 470...the above slides were at 460 FSB.

Wait till I tell the wife that

[BenchZowner]

was just looking at the read/wraite benches as a ratio , for example
My Bench Read 10317 Write 10126 Ratio 98.15% Your 10405/9791 = 94.10% Aboiut 400 points down.

I'm not sure i'm still a bit sceptical that the Everest write bench hasn't shifted, a bit like the old sandra timings slipped & changed through different versions.

Wait till I tell the wife that

Just a note:

The Write test is heavily dependent on the CPU's frequency

[Ace-a-Rue]

ok, thanks for that tidbit!

[Ace-a-Rue]

ok..i think my safe max FSB is 475 for the QX9650 (no surprise)...480 was not stable enough for video benches...it was reliable for easy load stuff.

[Supershanks]

The Write test is heavily dependent on the CPU's frequency

That could be the answer , on the Asus & early on with this board was using my Q6600 , rthen switched to the E8200 , that I got for the wife & she's not used yet But when I think back that's when I started seeing the pick up in write speed.

i am at the top end for my QX9650 in FSB...looks as if 480 is toooo much (can boot into windows but can't complete 3D2001)...trying a run at 475 FSB.

now, i did have the write speed come up since then as i climbed the FSB ladder to 470...the above slides were at 460 FSB.

Which falls in with what BenchZowner is saying I think.

Like we were saying the other day, dual core ,higher FSB gives more memory clocking options for sure with ddr3. Like I said then , max stable FSB on my Q6600 is 470. I've not done that much with my quad on this board.

[Ace-a-Rue]

Just a note:

The Write test is heavily dependent on the CPU's frequency

i have been doing some testing and it shows as you increase FSB so goes the "write" speed.

as i climbed the FSB at the same multi, like 8X, i raised my write speed to above 10K...once i increased my multi to 9X and then dropped the FSB back down to start off where i was at the same MHZ...the write speed dropped below 10K...SO...it appears that an increase in "write" speed is hinged to increasing FSB.

EDIT:...here you see 475FSB at 8X multi and write speed is above 10K...next slide is 450 FSB at 9X but higher Mhz than in 1st slide...again, you have lower write speed, below 10K

[BenchZowner]

Higher FSB helps.
Do a direct comparison... same FSB, same memset timings, just try with 8x multi and 9x

[Supershanks]

A few interesting things happening here:- 4,005 534x7.5

Do a direct comparison... same FSB, same memset timings, just try with 8x multi and 9x

Not particularly clear, at least, comparing :-
3,900 7.5x520Mhz @2.40B 5,5,5,12
to
4,160Mhz 8x520 520Mhz @2.40B 5,5,5,12

FSB..Multi...Read.....Write
520...7.5 ..10,664...10,895
520...8.0 ..10,814...10,989
% ...100.86%..101.41%

BUT FSB Coimparison seems to Highlight shift better:-
E8200 7.5x500 @3.33D = 1667DDR 7-6-5-18

E8200 7.5x510@1360Mhz DDR3
FSB..Multi...Read.....Write
510...7.5 ...9,813...10,278
520...7.5 ...9,723...10,528
101.96%. ..99.08%.102.43%

luck

[Ace-a-Rue]

thought to throw this up:

[Supershanks]

Nice going Ace What bios are you on ?? forgot to ask I'm using F5A

I'll be happy if the E8400 I have on order plays as well as this E8200


4,125 7.5x550@3.33D = 1,833Mhz 8,7,6,20 New high for FSB & also DDR3 rate at least at 1N


luck

[Ace-a-Rue]

thanks...this board is very good!

i am still on the original F4 with a date of 03/06/2008...i am thinking about trying F5A...how is it treating you?

[Ace-a-Rue]

A few interesting things happening here:- 4,005 534x7.5


Not particularly clear, at least, comparing :-
3,900 7.5x520Mhz @2.40B 5,5,5,12
to
4,160Mhz 8x520 520Mhz @2.40B 5,5,5,12

FSB..Multi...Read.....Write
520...7.5 ..10,664...10,895
520...8.0 ..10,814...10,989
% ...100.86%..101.41%

BUT FSB Coimparison seems to Highlight shift better:-
E8200 7.5x500 @3.33D = 1667DDR 7-6-5-18

E8200 7.5x510@1360Mhz DDR3
FSB..Multi...Read.....Write
510...7.5 ...9,813...10,278
520...7.5 ...9,723...10,528
101.96%. ..99.08%.102.43%

luck

any way you look at it, you are sure punching out some high read, write and copy speeds...this board is working great with your cellshock and it is working very well with my Gskill...i am very happy!

[Ace-a-Rue]

Nice going Ace What bios are you on ?? forgot to ask I'm using F5A

I'll be happy if the E8400 I have on order plays as well as this E8200


4,125 7.5x550@3.33D = 1,833Mhz 8,7,6,20 New high for FSB & also DDR3 rate at least at 1N


luck

here is another decent bench for Super-PI 32M:

[Ace-a-Rue]

kronos:

while i take some pics...look at these bios setting that SuperShanks posted up at his forum...these are for an E8200:

http://www.clunk.org.uk/forums/13663-post50.html

GIGABYTE GA X48T DQ6 Clock Settings

MB Intelligent Tweaker (MIT)
Robust Graphics Booster...........: Turbo
CPU Clock Ratio...................: 7
CPU Frequency.....................: 0.50
CPU Host Clock Control............: 3,750
CPU Host Frequency(Mhz)...........: 500
PCI Express Frequency (Mhz).......: 104
C.I.A.2...........................: Disabled
Performance Enhance...............: Turbo
System Memory Multiplier (SPD)....: 3.33D
Memory Frequency(Mhz).............: 1,667
DRAM Timing Selectable ...........: Manual
Standard Timing Control
CAS Latency Time (tCL)............: 7
DRAM RAS# to CAS#  (tRCD).........: 6
DRAM RAS# Precharge (tRP).........: 5
Precharge Delay (tRAS)............: 18
Advanced Timing Control
ACT to ACT Delay(tRRD)............: 5
Rank Write to READ Delay (tWTR)...: 5
Write To Precharge Delay (tWR)....: 10
Refresh to ACT Delay (tRFC).......: 60
Read To Precharge Delay (tRTP)....: 5
Static tREAD Value (tRD)..........: Auto
Static tREAD Phase Adj............: Auto
Command Rate (CMD)................: 1T
Clock Driving and Skew Control
CPU/PCIEX Clock Driving Control...: 900
CPU Clock Skew Control............:  Normal
(G)MCH Clock Skew Control.........: Normal
System Voltage Control () = Base Voltages
DDR3 OverVoltage Control...(1.50v): +0.35v
PCI-E OverVoltage Control.,(1.50v): +0.15v 
FSB OverVoltage Control....(1.10v): +0.20v
(G)MCH OverVoltage Control,(1.45v): +0.275v
CPU GTLREF1 Voltage*.......(67%)..: Normal
CPU GTLREF2 Voltage*.......(61%)..: Auto
Loadline Calibration..............: Disabled
CPU Voltage Control...............: 1.3125v
Normal CPU Vcore..................: 1.1127v
Note: * Hidden Features Press [CTRL] + [F1] from main Menu to Access
Advanced Bios Features
Limit CPUID Max to 3..............: Disabled
No-Execute Memory Protect.........:  Disabled
CPU Enhanced Halt(CIE)............: Disabled
CPU Thermal Monitor 2.............: Enabled (as yoiu want ) :)
CPU EIST Function.................: Disabled
Virtuaalization Technology........:  Disabled

[Ace-a-Rue]

first, upper bios screen of the X48T (1st installment):

middle of screen in 2nd slide.

last slide is self explanatory.

[kronos]

first, upper bios screen of the X48T (1st installment):

Thanks, Ace. I will try this later tonight.

[Ace-a-Rue]

more coming on previous page

[Ace-a-Rue]

one bench most folks like to see...this one is Prime 95 for one hour and ten minutes.

[Supershanks]

Wow, you've been busy since I went to bed :lol; great guide m8
Great SuperPi shots , ya blasted mine away

Like you said

.this board is working great with your cellshock and it is working very well with my Gskill...i am very happy!

It's not that often we get to say that, as early adopters of a new board & chipset.

i am still on the original F4 with a date of 03/06/2008...i am thinking about trying F5A...how is it treating you?

F5A is great no problems, though the only issue i have had was losing the Lan on the early F4H.

That's basically the only "issue" i've had.

Just going to plug my new E8400 in.

luck

[Ace-a-Rue]

you know there is someone else waiting around the corner to blast the next person out of the saddle...ALWAYS someone has a "better" bench to display...hehehe