P5B Deluxe timings 4-4-4 > 4-4-3 > 3-3-3 ?

[eva2000]

I just bought my first 965P chipset based board, Asus P5B Deluxe and playing around with E6600 B1 and 9x400fsb 1:1 boot from bios just to get a feel for the board.

I read that 4-4-4 is faster than 3-3-3 timings ????

I just tested 32M

400mhz 4-4-4-4 ~14min 39s untweaked and ~14min 32s memset tweaked
400mhz 4-4-3-4 ~14min 38s untweaked
400mhz 3-3-3-8 running right now but at half way mark is already 30 seconds slower than either 4-4-4-4 or 4-4-3-4 tests ????

I'm using 0804 bios.... anyone get to the bottom of this yet as to why 3-3-3-8 is so slow ?

thanks

[kiwi]

3-3-3 is slow for me too but 4-4-3 is better than 4-4-4. I am using 711 bios. Can you show your subtimings?

[eva2000]

Haven't uploaded all my results yet as still running 3-3-3-8 32M right now

but here's for 4-4-4-4 and 4-4-3-4 with untweaked OS and just bios set changes in timings NO memset changes at all

http://fileshosts.com/intel/Asus/P5B_Deluxe/results/G4Storm/E6600_1/ocz/titanium_alphavx2/pc8000/1_2/0804/9x/533/400-400-4444_2-20-9979_1.4375-auto-2.1/superpi-32m_single14m33s437s_tn.png (http://fileshosts.com/intel/Asus/P5B_Deluxe/results/G4Storm/E6600_1/ocz/titanium_alphavx2/pc8000/1_2/0804/9x/533/400-400-4444_2-20-9979_1.4375-auto-2.1/superpi-32m_single14m33s437s.png)

VS

http://fileshosts.com/intel/Asus/P5B_Deluxe/results/G4Storm/E6600_1/ocz/titanium_alphavx2/pc8000/1_2/0804/9x/533/400-400-4434_2-20-9979_1.4375-auto-2.1/superpi-32m_single14m38s281ms_tn.png (http://fileshosts.com/intel/Asus/P5B_Deluxe/results/G4Storm/E6600_1/ocz/titanium_alphavx2/pc8000/1_2/0804/9x/533/400-400-4434_2-20-9979_1.4375-auto-2.1/superpi-32m_single14m38s281ms.png)

looks like when i set tRP = 3, bios automatically changed All Precharge to Activate and All Precharge to Refresh delay from 4 to 3.

then tried 4-4-4-4 with memset tweak changed RAS to CAS Write delay from 4 to 2
http://fileshosts.com/intel/Asus/P5B_Deluxe/results/G4Storm/E6600_1/ocz/titanium_alphavx2/pc8000/1_2/0804/9x/533/400-400-4444_2-20-9979_1.4375-auto-2.1/memset/superpi-32m_single14m32s922s_tn.png (http://fileshosts.com/intel/Asus/P5B_Deluxe/results/G4Storm/E6600_1/ocz/titanium_alphavx2/pc8000/1_2/0804/9x/533/400-400-4444_2-20-9979_1.4375-auto-2.1/memset/superpi-32m_single14m32s922s.png)

[bachus_anonym]

Yep, welcome to the world of P5B Deluxe wackiness :D Useless of 3-3-3 is one of them :( I have done a little comparison somewhere a while ago - past "straight-booted" 400MHz FSB, 3-3-3 becomes faster than 4-4-4 but that's mainly due to chipset latencies being loosened up (or strap change, depending on who you ask :D), which super-oddly does not affect 3-3-3 at all, only 4-4-4 or 4-3-4-4 or 4-4-3 :)

http://www.xtremesystems.org/forums/showpost.php?p=1760236&postcount=28

[Viss]

I thought ras to cas 3 is the problem. 3-4-3-4 is faster for me then 3-3-3-4 but about the same as 4-4-4-4

http://img.photobucket.com/albums/v292/Viss1/3-3-3-48M.jpg
http://img.photobucket.com/albums/v292/Viss1/3-3-3-4Wrar.jpg

http://img.photobucket.com/albums/v292/Viss1/3-4-3-48M.jpg
http://img.photobucket.com/albums/v292/Viss1/3-4-3-4Wrar.jpg

http://img.photobucket.com/albums/v292/Viss1/4-4-4-48M.jpg
http://img.photobucket.com/albums/v292/Viss1/4-4-4-4Wrar.jpg

[Cappie]

Happens for me too, but found out that putting that one setting on 4 and rest to 3 fixed it indeed.

[eva2000]

thanks guys for the education! very weird stuff indeed

http://www.xtremesystems.org/forums/showpost.php?p=1760236&postcount=28

knew I could count on you to have comparison results handy :)

so would the reverse be true ? if i was to set 3-3-3-8 in bios and boot at 401+ fsb, the use clockgen to downclock to 400fsb and I should get times faster than boot at 400fsb 3-3-3-8 ?

[pentium777]

Am testing now my new rig BFG 680i 2x8800 GTX now that RAM speeds can be set really high I'm wondering will I benefit more running my TWIN2X-6400C4 @ 4-4-4-12 @ 900? Or so far I have been able to get 5-5-5-15 2.2v @ 1090 maybe 1100 depending on room temp ambient.

Basically I'm looking for better overall gaming / performance on the system either way.

Thanks

[shimmishim]

i think if you boot at 401 then you're on the 1333/1600 strap.

if you boot at 400, you're on the 1066 strap.

my guess would be 400 clockgen'ed down from 401 would be slower than straight boot at 400.

i get this board soon as well (today or tomorrow).

[eva2000]

pentium777 only way to know is test it for yourself i ;)

so far my tests concur with bachus and other folks findings

Bio Settings:

AI Tuning: Manual
CPU Frequency: 400
DRAM Frequency: DRR2-800Mhz
PCI Express Frequency: 100
PCI Clock Synchronization Mode: 33.33Mhz
Spread Spectrum: Disabled
Memory Voltage: 2.1v
CPU Vcore Voltage: 1.4375v (1.384v)
FSB Termination Voltage: AUTO
NB Vcore: AUTO
SB Vcore (SATA/PCIE): AUTO
ICH Chipset Voltage: AUTO


--------------------------------------------------
Straight boot from bios
@400mhz 4-4-4-4 2-20-3-9-7-9

Memory Remap Feature: Disabled
Configure DRAM Timing by SPD: Disabled
DDRAM CAS# Latency: 4
DRAM RAS# to CAS# Delay: 4
DRAM RAS# Precharge: 4
DRAM RAS# Activate to Precharge: 4
DRAM Write Recovery Time: 2
DRAM TRFC: 20
DRAM TRRD: 3
Rank Write to Read Delay: 9
Read to Precharge Delay: 7
Write to Precharge Delay: 9
Static Read Control: Disabled

http://fileshosts.com/intel/Asus/P5B_Deluxe/results/G4Storm/E6600_1/ocz/titanium_alphavx2/pc8000/1_2/0804/9x/533/400-400-4444_2-20-9979_1.4375-auto-2.1/superpi-32m_single14m33s437s_tn.png (http://fileshosts.com/intel/Asus/P5B_Deluxe/results/G4Storm/E6600_1/ocz/titanium_alphavx2/pc8000/1_2/0804/9x/533/400-400-4444_2-20-9979_1.4375-auto-2.1/superpi-32m_single14m33s437s.png)


--------------------------------------------------
Straight boot from bios
+
Memset tweak change RAS TO CAS Write Delay from 4 to 2
@400mhz 4-4-4-4 2-20-3-9-7-9

Memory Remap Feature: Disabled
Configure DRAM Timing by SPD: Disabled
DDRAM CAS# Latency: 4
DRAM RAS# to CAS# Delay: 4
DRAM RAS# Precharge: 4
DRAM RAS# Activate to Precharge: 4
DRAM Write Recovery Time: 2
DRAM TRFC: 20
DRAM TRRD: 3
Rank Write to Read Delay: 9
Read to Precharge Delay: 7
Write to Precharge Delay: 9
Static Read Control: Disabled

http://fileshosts.com/intel/Asus/P5B_Deluxe/results/G4Storm/E6600_1/ocz/titanium_alphavx2/pc8000/1_2/0804/9x/533/400-400-4444_2-20-9979_1.4375-auto-2.1/memset/superpi-32m_single14m32s922s_tn.png (http://fileshosts.com/intel/Asus/P5B_Deluxe/results/G4Storm/E6600_1/ocz/titanium_alphavx2/pc8000/1_2/0804/9x/533/400-400-4444_2-20-9979_1.4375-auto-2.1/memset/superpi-32m_single14m32s922s.png)


--------------------------------------------------
Straight boot from bios
@400mhz 4-4-3-4 2-20-3-9-7-9

Memory Remap Feature: Disabled
Configure DRAM Timing by SPD: Disabled
DDRAM CAS# Latency: 4
DRAM RAS# to CAS# Delay: 4
DRAM RAS# Precharge: 3
DRAM RAS# Activate to Precharge: 4
DRAM Write Recovery Time: 2
DRAM TRFC: 20
DRAM TRRD: 3
Rank Write to Read Delay: 9
Read to Precharge Delay: 7
Write to Precharge Delay: 9
Static Read Control: Disabled

http://fileshosts.com/intel/Asus/P5B_Deluxe/results/G4Storm/E6600_1/ocz/titanium_alphavx2/pc8000/1_2/0804/9x/533/400-400-4434_2-20-9979_1.4375-auto-2.1/superpi-32m_single14m38s281ms_tn.png (http://fileshosts.com/intel/Asus/P5B_Deluxe/results/G4Storm/E6600_1/ocz/titanium_alphavx2/pc8000/1_2/0804/9x/533/400-400-4434_2-20-9979_1.4375-auto-2.1/superpi-32m_single14m38s281ms.png)



--------------------------------------------------
Straight boot from bios
@400mhz 3-3-3-8 6-42-10-10-10-10


Bio Settings:

AI Tuning: Manual
CPU Frequency: 400
DRAM Frequency: DRR2-800Mhz
PCI Express Frequency: 100
PCI Clock Synchronization Mode: 33.33Mhz
Spread Spectrum: Disabled
Memory Voltage: 2.25v
CPU Vcore Voltage: 1.4375v (1.384v)
FSB Termination Voltage: AUTO
NB Vcore: AUTO
SB Vcore (SATA/PCIE): AUTO
ICH Chipset Voltage: AUTO

Memory Remap Feature: Disabled
Configure DRAM Timing by SPD: Disabled
DDRAM CAS# Latency: 3
DRAM RAS# to CAS# Delay: 3
DRAM RAS# Precharge: 3
DRAM RAS# Activate to Precharge: 8
DRAM Write Recovery Time: 6
DRAM TRFC: 42
DRAM TRRD: 10
Rank Write to Read Delay: 10
Read to Precharge Delay: 10
Write to Precharge Delay: 10
Static Read Control: Disabled

http://fileshosts.com/intel/Asus/P5B_Deluxe/results/G4Storm/E6600_1/ocz/titanium_alphavx2/pc8000/1_2/0804/9x/533/400-400-3338_def_1.4375-auto-2.1/superpi-32m_single15m35s734s_tn.png (http://fileshosts.com/intel/Asus/P5B_Deluxe/results/G4Storm/E6600_1/ocz/titanium_alphavx2/pc8000/1_2/0804/9x/533/400-400-3338_def_1.4375-auto-2.1/superpi-32m_single15m35s734s.png)


--------------------------------------------------
Straight boot from bios
@400mhz 3-3-3-8 2-20-3-9-7-9

Bio Settings:

AI Tuning: Manual
CPU Frequency: 400
DRAM Frequency: DRR2-800Mhz
PCI Express Frequency: 100
PCI Clock Synchronization Mode: 33.33Mhz
Spread Spectrum: Disabled
Memory Voltage: 2.25v
CPU Vcore Voltage: 1.4375v (1.384v)
FSB Termination Voltage: AUTO
NB Vcore: AUTO
SB Vcore (SATA/PCIE): AUTO
ICH Chipset Voltage: AUTO

Memory Remap Feature: Disabled
Configure DRAM Timing by SPD: Disabled
DDRAM CAS# Latency: 3
DRAM RAS# to CAS# Delay: 3
DRAM RAS# Precharge: 3
DRAM RAS# Activate to Precharge: 8
DRAM Write Recovery Time: 2
DRAM TRFC: 20
DRAM TRRD: 3
Rank Write to Read Delay: 9
Read to Precharge Delay: 7
Write to Precharge Delay: 9
Static Read Control: Disabled

http://fileshosts.com/intel/Asus/P5B_Deluxe/results/G4Storm/E6600_1/ocz/titanium_alphavx2/pc8000/1_2/0804/9x/533/400-400-3338_2-20-9979_1.4375-auto-2.1/superpi-32m_single15m24s125s_tn.png (http://fileshosts.com/intel/Asus/P5B_Deluxe/results/G4Storm/E6600_1/ocz/titanium_alphavx2/pc8000/1_2/0804/9x/533/400-400-3338_2-20-9979_1.4375-auto-2.1/superpi-32m_single15m24s125s.png)


--------------------------------------------------
Straight boot from bios
@402mhz 3-3-3-8 2-20-3-9-7-9
+
Clockgen down to 400mhz 3-3-3-8

Bio Settings:

AI Tuning: Manual
CPU Frequency: 402
DRAM Frequency: DRR2-804Mhz
PCI Express Frequency: 100
PCI Clock Synchronization Mode: 33.33Mhz
Spread Spectrum: Disabled
Memory Voltage: 2.25v
CPU Vcore Voltage: 1.4375v (1.384v)
FSB Termination Voltage: AUTO
NB Vcore: AUTO
SB Vcore (SATA/PCIE): AUTO
ICH Chipset Voltage: AUTO

Memory Remap Feature: Disabled
Configure DRAM Timing by SPD: Disabled
DDRAM CAS# Latency: 3
DRAM RAS# to CAS# Delay: 3
DRAM RAS# Precharge: 3
DRAM RAS# Activate to Precharge: 8
DRAM Write Recovery Time: 2
DRAM TRFC: 20
DRAM TRRD: 3
Rank Write to Read Delay: 9
Read to Precharge Delay: 7
Write to Precharge Delay: 9
Static Read Control: Disabled

http://fileshosts.com/intel/Asus/P5B_Deluxe/results/G4Storm/E6600_1/ocz/titanium_alphavx2/pc8000/1_2/0804/9x/533/402-402-3338_2-20-9979_1.4375-auto-2.1/clockgen_400fsb/superpi-32m_single15m02s515s_tn.png (http://fileshosts.com/intel/Asus/P5B_Deluxe/results/G4Storm/E6600_1/ocz/titanium_alphavx2/pc8000/1_2/0804/9x/533/402-402-3338_2-20-9979_1.4375-auto-2.1/clockgen_400fsb/superpi-32m_single15m02s515s.png)


--------------------------------------------------
Straight boot from bios
@400mhz 3-4-3-4 2-20-3-9-7-9

Bio Settings:

AI Tuning: Manual
CPU Frequency: 400
DRAM Frequency: DRR2-800Mhz
PCI Express Frequency: 100
PCI Clock Synchronization Mode: 33.33Mhz
Spread Spectrum: Disabled
Memory Voltage: 2.25v
CPU Vcore Voltage: 1.4375v (1.384v)
FSB Termination Voltage: AUTO
NB Vcore: AUTO
SB Vcore (SATA/PCIE): AUTO
ICH Chipset Voltage: AUTO

Memory Remap Feature: Disabled
Configure DRAM Timing by SPD: Disabled
DDRAM CAS# Latency: 3
DRAM RAS# to CAS# Delay: 4
DRAM RAS# Precharge: 3
DRAM RAS# Activate to Precharge: 4
DRAM Write Recovery Time: 2
DRAM TRFC: 20
DRAM TRRD: 3
Rank Write to Read Delay: 9
Read to Precharge Delay: 7
Write to Precharge Delay: 9
Static Read Control: Disabled

http://fileshosts.com/intel/Asus/P5B_Deluxe/results/G4Storm/E6600_1/ocz/titanium_alphavx2/pc8000/1_2/0804/9x/533/400-400-3434_2-20-3-9979_1.4375-auto-2.25/superpi-32m_single_14m34s031ms_tn.png (http://fileshosts.com/intel/Asus/P5B_Deluxe/results/G4Storm/E6600_1/ocz/titanium_alphavx2/pc8000/1_2/0804/9x/533/400-400-3434_2-20-3-9979_1.4375-auto-2.25/superpi-32m_single_14m34s031ms.png)


--------------------------------------------------
Straight boot from bios
@400mhz 3-4-3-4 2-20-3-9-7-9
+
Memset tweak change RAS TO CAS Write Delay from 4 to 2

Bio Settings:

AI Tuning: Manual
CPU Frequency: 400
DRAM Frequency: DRR2-800Mhz
PCI Express Frequency: 100
PCI Clock Synchronization Mode: 33.33Mhz
Spread Spectrum: Disabled
Memory Voltage: 2.25v
CPU Vcore Voltage: 1.4375v (1.384v)
FSB Termination Voltage: AUTO
NB Vcore: AUTO
SB Vcore (SATA/PCIE): AUTO
ICH Chipset Voltage: AUTO

Memory Remap Feature: Disabled
Configure DRAM Timing by SPD: Disabled
DDRAM CAS# Latency: 3
DRAM RAS# to CAS# Delay: 4
DRAM RAS# Precharge: 3
DRAM RAS# Activate to Precharge: 4
DRAM Write Recovery Time: 2
DRAM TRFC: 20
DRAM TRRD: 3
Rank Write to Read Delay: 9
Read to Precharge Delay: 7
Write to Precharge Delay: 9
Static Read Control: Disabled

http://fileshosts.com/intel/Asus/P5B_Deluxe/results/G4Storm/E6600_1/ocz/titanium_alphavx2/pc8000/1_2/0804/9x/533/400-400-3434_2-20-3-9979_1.4375-auto-2.25/memset/superpi-32m_single_14m29s156ms_tn.png (http://fileshosts.com/intel/Asus/P5B_Deluxe/results/G4Storm/E6600_1/ocz/titanium_alphavx2/pc8000/1_2/0804/9x/533/400-400-3434_2-20-3-9979_1.4375-auto-2.25/memset/superpi-32m_single_14m29s156ms.png)

--------------------------------------------------
Really weird behaviour going on here! Seems below 400FSB 4-4-4 and 4-4-3 timings are faster than 3-3-3 and the culprit is tRCD value, but above 401FSB, tRCD 3 (i.e. 3-3-3) picks up performance like expected it seems if you read http://www.xtremesystems.org/forums/showpost.php?p=1760236&postcount=28 and http://www.xtremesystems.org/forums/showthread.php?t=123284

[shimmishim]

WHOA

interesting...

straight boot 400 vs. boot up at 402 and clockgen down is interesting indeed!

at cas 3 it's faster when you clockgen down. that is definitely a surprise!

thanks for the results.

[bachus_anonym]

Nice work :)

If you want to play more, boot @ 400 8-4-4-4.0 and run e.g. 32M. Then after it's done, use MemSet to "tighten" to 8-3-3-4.0 (or anything else with TRCD=3) and run 32M again... Watch how 32M result face-plants. I tried it few times and this what happened - i might have a screen somewhere... Funny, 3-3-3 slower than 4-4-4, eh :lol: That's pretty much the only beef I have with ASUS and this board...

Your example of boot @ 402 and down to 400 is quite interesting... I never tried that. Though, it's still slower than boot @ 400 4-4-4...

[thorgal]

Great job again, wow :toast:

Come to think of it, I'll have to be carefull to post results with cas 3 in my upcoming reviews...

[erwinz]

nice.. :) really thinking about geting this board.. ;)

[eva2000]

thanks guys
Nice work :)

If you want to play more, boot @ 400 8-4-4-4.0 and run e.g. 32M. Then after it's done, use MemSet to "tighten" to 8-3-3-4.0 (or anything else with TRCD=3) and run 32M again... Watch how 32M result face-plants. I tried it few times and this what happened - i might have a screen somewhere... not sure i follow ? you mean setting timings using memset is slower than straight boot from bios set timings ?

[sskings]

i think if you boot at 401 then you're on the 1333/1600 strap.

if you boot at 400, you're on the 1066 strap.

my guess would be 400 clockgen'ed down from 401 would be slower than straight boot at 400.

i get this board soon as well (today or tomorrow).


Let us know if your chipset is c1 or c2. I've heard there differences with memory timings.

Thanks.

[IluvIntel]

Can someone please try 3-2-2-4 @ 400Mhz, probably only those with Vdimm mods or extremely nice DDR2 !
I know exactly what everyones talking about with P965 chipset & 32Mb SuperPi...:D and I thought my board or hardware combo was crap...

[Will Eatforfood]

I dont think trying those timings are necessary. 3-3-3-x didn't do so well so what would setting it tighter do? We will see the same trend.

[Smash]

Nice Tests:clap:

And what's when you boot from 400 or 402 and go then on overclocking? Are then the lower timings better?

[over@locker886]

Thank you great test eva!

[rjw]

good job

[deadlock7]

Actually ther best timings is 4-4-2 for 1066 strap.Even 3-2-2 is worst..

[Revv23]

man ram clocks so nice in the P5B it's too bad they don't give us proper control of the timings.

[eva2000]

Actually ther best timings is 4-4-2 for 1066 strap.Even 3-2-2 is worst..
interesting haven't tried 4-4-2 yet :)

[IluvIntel]

I dont think trying those timings are necessary. 3-3-3-x didn't do so well so what would setting it tighter do? We will see the same trend.

It could be logical to presume that, but it does'nt mean that will be the case. Has someone tried this yet ?

[deadlock7]

man ram clocks so nice in the P5B it's too bad they don't give us proper control of the timings.

Ram clocks better than 975,but i dislike 4-4-2-8!:mad:


interesting haven't tried 4-4-2 yet :)

It's just a bit faster than 4-4-3.;)

[hipro5]

May I contribute?....:D

A friend brought me one for Moding it, so I "play" a bit with it......:D

RULE No1: DON'T use 1:1 devider on this mobo.....It's NOT 975 chipset....;)

Take a GOOD look at the screenies below AND the LATENCY.....;)

http://www.thelab.gr/images/Hipro5/CONROE_X6800ES/P5B-Deluxe/Everest_Latency_1.png

http://www.thelab.gr/images/Hipro5/CONROE_X6800ES/P5B-Deluxe/Everest_Latency_2.png

http://www.thelab.gr/images/Hipro5/CONROE_X6800ES/P5B-Deluxe/Everest_Latency_3.png

http://www.thelab.gr/images/Hipro5/CONROE_X6800ES/P5B-Deluxe/Everest_Latency_4.png

http://www.thelab.gr/images/Hipro5/CONROE_X6800ES/P5B-Deluxe/Everest_Latency_5.png

http://www.thelab.gr/images/Hipro5/CONROE_X6800ES/P5B-Deluxe/Everest_Latency_6.png

http://www.thelab.gr/images/Hipro5/CONROE_X6800ES/P5B-Deluxe/Everest_Latency_7.png

http://www.thelab.gr/images/Hipro5/CONROE_X6800ES/P5B-Deluxe/Everest_Latency_8.png

.....and AGAIN......

http://www.thelab.gr/images/Hipro5/CONROE_X6800ES/P5B-Deluxe/Everest_Latency_9.png


AS you can see, the BEST is at 4-4-4-x ......and with 4:5 OR 2:3......1:1 gives a very high latency 55 - 60ns......

http://www.thelab.gr/images/Hipro5/CONROE_X6800ES/P5B-Deluxe/Super-Pi_32M_3604_ram_563_13m03s750ms_1.png

http://www.thelab.gr/images/Hipro5/CONROE_X6800ES/P5B-Deluxe/Super-Pi_32M_3604_ram_563_13m03s750ms_2.png

http://www.thelab.gr/images/Hipro5/CONROE_X6800ES/P5B-Deluxe/Super-Pi_1M_3600_ram_600_13.906_1.png

http://www.thelab.gr/images/Hipro5/CONROE_X6800ES/P5B-Deluxe/Super-Pi_1M_3600_ram_600_13.906_2.png

http://www.thelab.gr/images/Hipro5/CONROE_X6800ES/P5B-Deluxe/Super-Pi_1M_3604_ram_563_13.891s.png


.

[Revv23]

May I contribute?....:D

A friend brought me one for Moding it, so I "play" a bit with it......:D

RULE No1: DON'T use 1:1 devider on this mobo.....It's NOT 975 chipset....;)

Take a GOOD look at the screenies below AND the LATENCY.....;)

AS you can see, the BEST is at 4-4-4-x ......and with 4:5 OR 2:3......1:1 gives a very high latency 55 - 60ns......


.

Wow hipro thats a big help thanks!

What mods are you doing to it if i may ask?

edit - Also, a question, did you switch timings via memset or in bios? as bios timings seem to effect performance different then memset.

[Viss]

450 4:5 is nice. What settings/bios you used for that?

[hipro5]

I booted at 400MHz SHARP so as to keep a better strap with Cas 5-4-4-8 -6-5-10-10-10-12-disable.....I then within Windows, uped the fsb at 450MHz and played with the Ram's timmings/subtimmings....Doing so, I had a 43.2ns latency on everest and 8ns latency on S-Sandra.....Also 2007 score on winrar 3.61 benchmark.....:)
Winrar benchmark shows you what you've got...;)


Wow hipro thats a big help thanks!

What mods are you doing to it if i may ask?

edit - Also, a question, did you switch timings via memset or in bios? as bios timings seem to effect performance different then memset.

Mods - simple things - Only VMCH and VICH-VTT....:)

@ eva2000: Put the tRRD=5.....On this mobo, tighter is not better...... ;)

[funkflix]

Thx for all the nice tests here! :toast:

Anyone have a tip for me, why i can't use the 4:5 divider? I can not boot into windows, doesn't matter which timings or bios i use.. :confused: 9x400 and my Fatbodies @ 500MHz and with 2,4V Vdimm = no go!

With my E6400 i already handled to bench with 505MHz, but it was 1:1 and with "strange" Timings. http://img243.imageshack.us/img243/6509/5053328smalldr3.jpg

[ChaosMinionX]

hipro is that a 6800 running 400x9 at 4:5 ddr1000? i have been trying forever to get that is that stable or only benchable?

[hipro5]

Thx for all the nice tests here! :toast:

Anyone have a tip for me, why i can't use the 4:5 divider? I can not boot into windows, doesn't matter which timings or bios i use.. :confused: 9x400 and my Fatbodies @ 500MHz and with 2,4V Vdimm = no go!

With my E6400 i already handled to bench with 505MHz, but it was 1:1 and with "strange" Timings. http://img243.imageshack.us/img243/6509/5053328smalldr3.jpg

Fatbodies and 4:5 = 5-4-4-12-6-10-10-10-12-disable

[hipro5]

hipro is that a 6800 running 400x9 at 4:5 ddr1000? i have been trying forever to get that is that stable or only benchable?

Yes it's an X6800 one....Vdimm = 2.70Volts via DDR Maximizer as for the Corsair 8000UL Fatbodies and 2.66Volts as for the Cell-Shock 1000C4 ones...It benchies everything ....:)

P.S. Try PCI Latency=128 and PCI-E = 110 to 120MHz.....

[funkflix]

Fatbodies and 4:5 = 5-4-4-12-6-10-10-10-12-disable

Thx mate, but it have not worked(i forgot the 12 on the end to be honest, it was on 10). Are this the tightest timings possible with Fatbodies on 4:5 or just the timings to be sure, that it is working 100%?

Will give it a try again tomorrow, with a little bit more sleep and time.. ;)

[hipro5]

Thx mate, but it have not worked(i forgot the 12 on the end to be honest, it was on 10). Are this the tightest timings possible with Fatbodies on 4:5 or just the timings to be sure, that it is working 100%?

Will give it a try again tomorrow, with a little bit more sleep and time.. ;)

You can pick 5-5-5-15 as well for boot....

[clon22]

I booted at 400MHz SHARP so as to keep a better strap with Cas 5-4-4-8 -6-5-10-10-10-12-disable.....I then within Windows, uped the fsb at 450MHz and played with the Ram's timmings/subtimmings....Doing so, I had a 43.2ns latency on everest and 8ns latency on S-Sandra.....Also 2007 score on winrar 3.61 benchmark.....:)
Winrar benchmark shows you what you've got...;)



Mods - simple things - Only VMCH and VICH-VTT....:)

@ eva2000: Put the tRRD=5.....On this mobo, tighter is not better...... ;)
Wow!...
Thanks for all the effort!..
Hmm.. the sub timings r very interesting...
I'm running my 6403 @ 3-4-3-4-3-20-5-9-6-9
gonna try out ur tests...
Thou my board is only a P5Bvanilla.. :(

[funkflix]

You can pick 5-5-5-15 as well for boot....

Tested that all before, but can't get it working here, thanks anyway..

[v_parrello]

Very interesting thread guys! Thanks a lot!

I have two questions for you:
- is it true for all 965 mobos?
- what about mobos with 975 chipset? They perform in the same way?

Tanks in advance for your answers.

[eva2000]

@ eva2000: Put the tRRD=5.....On this mobo, tighter is not better...... ;)
thanks for the tip mate :)

[»X$«êdên]

:rocker: YAY thank you!
I have this Memory:OCZ Platinum Revision, 2 2GB, (PC2 6400), Rated @ 4-4-4-15; and it seems to like that at 266 if EVER...
@333 1:1 my core temp is 62 / 58 since the motherboard switch with added fans. :(
I need to get that everest!

[deadlock7]

Hmm..i was playing with sandra and 4-4-2 gave me better results than 4-4-4.I tried everest and spi and i found out that Hipro is 100% right.Thanks for the tips dude..

[Millyons]

for me out of all of them 3-4-4-x was the best for PI 1mb

[cpulloverclock]

the best timings for high frequency memory with a P5B with is 3-4-2-4-4-3-20-1 :)

the problem about 3-3-3 and 3-4-4-3 is known, it's why I didn't use them when I benched :o

just a P5B issue

[Revv23]

I booted at 400MHz SHARP so as to keep a better strap with Cas 5-4-4-8 -6-5-10-10-10-12-disable.....I then within Windows, uped the fsb at 450MHz and played with the Ram's timmings/subtimmings....Doing so, I had a 43.2ns latency on everest and 8ns latency on S-Sandra.....Also 2007 score on winrar 3.61 benchmark.....:)
Winrar benchmark shows you what you've got...;)



Mods - simple things - Only VMCH and VICH-VTT....:)

@ eva2000: Put the tRRD=5.....On this mobo, tighter is not better...... ;)


Cool thanks again hipro, any chance you havea picture of that VICH-VTT mod, ive never seen it.

[hipro5]

http://www.thelab.gr/images/Hipro5/P5B_Deluxe/VICH_VTT_Mod_1.jpg

http://www.thelab.gr/images/Hipro5/P5B_Deluxe/VICH_VTT_Mod_2.jpg

http://www.thelab.gr/images/Hipro5/P5B_Deluxe/Vdroop_Mod.jpg

:)

[-n7-]

Results from a little testing i did:

For these quick benches, i ran:
7x401 (2807 MHz) 1:1 DDR2-802 (4-4-3-4)
or
7x401 (2807 MHz) 1:1 DDR2-802 (3-3-3-4)
or
7x401 (2807 MHz) 4:5 DDR2-1002.5 (4-4-4-4)

Results were averaged from a few runs

SuperPI 4M
1:1 (4-4-3-4) - 104.875s
1:1 (3-3-3-4) - 103.964s
4:5 (4-4-4-4) - 101.990s

3DMark01 SE game tests only
1:1 (4-4-3-4) - 37653
1:1 (3-3-3-4) - 38040
4:5 (4-4-4-4) - 38645.5

Q4 bench 800x600 high quality AA/AF off SMP enabled
1:1 (4-4-3-4) - 97.91 fps
1:1 (3-3-3-4) - 99.53 fps
4:5 (4-4-4-4) - 102.4 fps

WinRAR bench multithreaded
1:1 (4-4-3-4) - 1184 KB/s
1:1 (3-3-3-4) - 1225 KB/s
4:5 (4-4-4-4) - 1326 KB/s

FEAR 800x600 0x AA 2x AF everything maxed
Min - Ave. - Max fps
1:1 (4-4-3-4) - 55 / 111.67 / 231.67
1:1 (3-3-3-4) - 55 / 111 / 229.33
4:5 (4-4-4-4) - 55.33 / 112.67 / 233.33

UT2k4 Primeval 1024x768 no AA/AF
1:1 (4-4-3-4) - 91.965 fps
1:1 (3-3-3-4) - 92.975 fps
4:5 (4-4-4-4) - 94.635 fps


Those results indicate CAS 3 works just fine for me...:confused:

[pentium777]

Thank you for all that testing although I am on 680i I'd still like to ensure I am not missing anything? running that 4:5 and looser timings but higher DDR2 speeds provided much better results in bench, gaming and everything you did. So why would I want to run 3-3-3 if I can loosen and run higher DDR2 speed and provide better overall system performance?

[Millyons]

Results from a little testing i did:

For these quick benches, i ran:
7x401 (2807 MHz) 1:1 DDR2-802 (4-4-3-4)
or
7x401 (2807 MHz) 1:1 DDR2-802 (3-3-3-4)
or
7x401 (2807 MHz) 4:5 DDR2-1002.5 (4-4-4-4)

Results were averaged from a few runs

SuperPI 4M
1:1 (4-4-3-4) - 104.875s
1:1 (3-3-3-4) - 103.964s
4:5 (4-4-4-4) - 101.990s

3DMark01 SE game tests only
1:1 (4-4-3-4) - 37653
1:1 (3-3-3-4) - 38040
4:5 (4-4-4-4) - 38645.5

Q4 bench 800x600 high quality AA/AF off SMP enabled
1:1 (4-4-3-4) - 97.91 fps
1:1 (3-3-3-4) - 99.53 fps
4:5 (4-4-4-4) - 102.4 fps

WinRAR bench multithreaded
1:1 (4-4-3-4) - 1184 KB/s
1:1 (3-3-3-4) - 1225 KB/s
4:5 (4-4-4-4) - 1326 KB/s

FEAR 800x600 0x AA 2x AF everything maxed
Min - Ave. - Max fps
1:1 (4-4-3-4) - 55 / 111.67 / 231.67
1:1 (3-3-3-4) - 55 / 111 / 229.33
4:5 (4-4-4-4) - 55.33 / 112.67 / 233.33

UT2k4 Primeval 1024x768 no AA/AF
1:1 (4-4-3-4) - 91.965 fps
1:1 (3-3-3-4) - 92.975 fps
4:5 (4-4-4-4) - 94.635 fps


Those results indicate CAS 3 works just fine for me...:confused:

What about when you boot at 400mhz

[alexio]

Has anybody actually tried to hardcode tighter timings in the SPD? I think doing this and booting with SPD settings should fix the problem.

What do you guys think?

[meloxicam]

I am having trouble with Clockgen crashing with 2 sticks of ram on PB5.

I am booting at 400 which helps my latency on my PB5 by over 15ns to 49ns. But when I use Clockgen 1.0.5.1 to increase to 425FSB, which is the best speed in the BIOS that my e6300 will do completely stable, the system blue screens. Funny thing is that if I take out 1 of my Crucial 1gig Anv. editions, it runs fine with Clockgen @ 425.

I have tried a different set of ram too, and it still gives same results. Any idea why Clockgen would have a problem with 2x1gig, yet run fine with one stick?

I am running the ram at 4:5 and I am aware that sometimes with 2 sticks you need more voltage, but it's completely stable if I boot 425 in the BIOS. This sucks because with 425 in the BIOS my Everest latency is 65ns. I wish they could fix the latency issues with the BIOS on the PB5. :(

[Viss]

Youll probably need a modded board to clockgen up over 425~430 booting at 400. 4/5 divider makes it even harder. Read this (http://www.xtremesystems.org/forums/showthread.php?t=114998) tread if you havent already.

[dinos22]

Nice work :)

If you want to play more, boot @ 400 8-4-4-4.0 and run e.g. 32M. Then after it's done, use MemSet to "tighten" to 8-3-3-4.0 (or anything else with TRCD=3) and run 32M again... Watch how 32M result face-plants. I tried it few times and this what happened - i might have a screen somewhere... Funny, 3-3-3 slower than 4-4-4, eh :lol: That's pretty much the only beef I have with ASUS and this board...

Your example of boot @ 402 and down to 400 is quite interesting... I never tried that. Though, it's still slower than boot @ 400 4-4-4...
what about dodgy 4:5 divider :( :slapass:

[meloxicam]

Youll probably need a modded board to clockgen up over 425~430 booting at 400. 4/5 divider makes it even harder. Read this tread if you havent already.

Thanks, I have read the thread. From every thing I have seen, 4:5 seems to perform considerably better on PB5.

Does Clockgen have an issue with 4:5?

What MB mod is used for Clockgen that you mentioned?

[Viss]

If you want to stay on fast strap and CG up higher youll need vmods done to the board. Its not a 4/5 divider clockgen problem, i dont think youll get much higher using 1/1.

[eva2000]

Thanks, I have read the thread. From every thing I have seen, 4:5 seems to perform considerably better on PB5.

Does Clockgen have an issue with 4:5?

What MB mod is used for Clockgen that you mentioned?
yeah NB bolts need volt modding for higher CG from 400 boot up past 430FSB

[Elisha]

i wish the P5B had a lower divider for the ram like the 875/865.
my cheapy crucial rendition can do 450mhz but the cpu can do 500fsb.
is there any way for forcing a lower divider for ram?

[meloxicam]

yeah NB bolts need volt modding for higher CG from 400 boot up past 430FSB

I am only looking to hit 425 and Clockgen works fine with only one stick of ram.

Any idea why it works fine with one stick and not 2?

[alexio]

i wish the P5B had a lower divider for the ram like the 875/865.
my cheapy crucial rendition can do 450mhz but the cpu can do 500fsb.
is there any way for forcing a lower divider for ram?
You could flash a DDR2-480 SPD and set memory speed to auto in the bios.

500 / (266.67/240) = 450mhz

[Elisha]

You could flash a DDR2-480 SPD and set memory speed to auto in the bios.

500 / (266.67/240) = 450mhz


how do i do that? flash with what?>
what happens if i just select Auto? would it run SPD or 1:1?

[alexio]

how do i do that? flash with what?>
what happens if i just select Auto? would it run SPD or 1:1?
It should run SPD.

Please read this: http://www.xtremesystems.org/forums/showthread.php?t=123284&page=3

Read and save your SPD from one of the two moduls and send it to my email:
alexkok33 4T hotmail D0T com
I'll see what I can do. You can rewrite SPD afterwards with SPD tool. If you want any custom timings, etc, PM me or do it yourself.

[Elisha]

i emailed you.
i would love to be able to have a 5:4 divider instead.

[Revv23]

965 doesnt run lower then 1:1 IIRC, no flashing of SPD will help.


what about dodgy 4:5 divider :( :slapass:

Last i checked the P5B was the best clocking 965 in 4:5 mode that you can buy. i would hardly call that dodgy.

[alexio]

965 doesnt run lower then 1:1 IIRC, no flashing of SPD will help.
So you mean a Conroe or Allendale can't run with DDR2-400 on stock setting? Does it force the DDR2-400 memory to DDR2-533 then, aka 1:1?

[JoeBar]

Yes the lowest setting is 1/1.

[alexio]

Yes the lowest setting is 1/1.
Alright. So this would mean that upcoming Conroes and Kentsfields with 333mhz FSB won't work with DDR2-533?

It's very important for me to know if the P5B Deluxe will force DDR2-400 to DDR2-533, or if it will not boot at all. Does anybody here know the answer for sure?

[Yoxxy]

It won't boot with DDR2-400. Needs to see 533 mhz PC2-4300 or higher.

[Viss]

Posted last night in another tread but thought it would be better of here in this one. In the link on first page there’s the thought that when using trcd 3 the strap doesn’t change at 401 as it does with 4-4-4 for example. Sandra BW does get lower at 9x401 3-3-3-4 though but not as much compared to 4-4-4 or 3-4-3. Winrar bench gets much better at 9x401 with 3-3-3-4. Whats goin on :confused:

9x400
3-4-3-4 = 1530 / 7422
4-4-4-4 = 1520 / 7392
3-3-3-4 = 1297 / 7298

9x401
3-3-3-4 = 1428 / 7033
3-4-3-4 = 1403 / 7002
4-4-4-4 = 1373 / 6903

3-3-3-4
http://img.photobucket.com/albums/v636/loc_oc/4003-3-3-4.jpg

http://img.photobucket.com/albums/v636/loc_oc/4013-3-3-4.jpg

3-4-3-4
http://img.photobucket.com/albums/v636/loc_oc/4003-4-3-4.jpg

http://img.photobucket.com/albums/v636/loc_oc/3-4-3-4.jpg

4-4-4-4
http://img.photobucket.com/albums/v636/loc_oc/4004-4-4-4.jpg

http://img.photobucket.com/albums/v636/loc_oc/4014-4-4-4.jpg

[funkflix]

That 401MHz is straight out of the bios or clockgened? Maybe u can try to compare that settings with Hexus Pifast too? I always prefer it for comparing timings and that stuff, but don't know if it affects the same way it does on A64.

[alexio]

Very interesting result. This means RtCD is somehow giving a bug at the 1067 strap. Can you test these values at 9*400mhz for full reference:
4-3-3-4
4-3-4-4
4-4-3-4
3-3-4-4
3-4-4-4
3-5-3-4
3-6-3-4

thanks ;)

[dinos22]

try see how far past 400MHz you can go too

[Viss]

Should have mentioned, booted at 9x400 and 401 from bios, no clockgen used.


try see how far past 400MHz you can go too

I can get up to 435 with clockgen if boot at 400 or less. I think its about avarage for unmodded board.