**The ASrock 939dual-SATA2 (ULi M1695) thread**

Hi, has anyone got a pic of this board with a freezer 64 pro on it?? I can't seem to get mine to fit, one of the board's heatsinks gets in the way!!
 
ScoTTyBEEE said:
Hi, has anyone got a pic of this board with a freezer 64 pro on it?? I can't seem to get mine to fit, one of the board's heatsinks gets in the way!!
Click to expand...

It does fit but it is very close to the heatsink, I'll get a picture if you want just waiting till the batteries charge up a bit
 
mp260767 said:
barnsta:

drop the htt to 600 and set the memory divider to 133, then see if you can up the fsb.

disco_stu:

depends how much u plan to overclock the ocw 2/3 bios' may have been better as they allow over 275fsb, otherwise you should be fine with the 140.
Click to expand...


sorry for newbieness, can't find the htt setting in bios? can u explain about mem divider please.

Thanks
 
Here's a really brief guide:

You have a base speed setting, from which almost everything is set. This is often called the HTT (which is also used to refer to the two hypertransport links) or the FSB, though neither term is really right. FSB is probably the better term, though it isn't the same as the FSB on other x86 architectures. Asrock call it 'CPU frequency' in the BIOS.

Anyway, this is the key setting. Unless the BIOS has changed a lot (I'm running P1.20), you'll find it in the Advanced menu screen in the BIOS. Set the overclock mode to CPU/PCIE async. If you don't, any overclocking will overclock the PCI Express bus and it won't like that. The base speed setting, whatever you want to call it, is below.

You will also need to change some other settings, because they will be set relative to the base speed.

Firstly, the memory thing. By default, the memory runs at the base speed (EDIT: assuming you are using PC3200), so if you increase that in order to overclock your CPU, you will overlock your memory by the same proportion. That will limit your overclocking unless you have unusually fast memory. You will probably be able to o/c your CPU by 25% by keeping the same CPU clock multiplier and increasing the base speed to 250MHz, but it's very unlikely that your memory will be able to overclock that much. For PC3200, it would be a 25% overclock. So you will need to set the memory to run at a fraction of the base speed - this is the memory divider.

An example: Say you have PC3200 memory, which is rated at 200MHz, and you set the base speed to 250MHz to get a 25% o/c on your CPU. In order to keep your memory at its rated speed, you have to run the memory at 4/5ths of the base speed - a memory divider of 4/5. The BIOS on this board is a little limited with memory dividers, as there are only three settings. You can have 1:1, 5/6ths or 2/3rds and that's all. So in this example you can set your memory to 250MHz, 208 1/3MHz (5/6ths) or 166 2/3MHz (2/3rds). The memory dividers are wrongly labelled in this BIOS. 1:1 is labelled as '200MHz', 5/6ths is labelled as '166MHz' and 2/3rds is labelled as '133MHz'. These labels are wrong and misleading.

There are two Hypertransport settings in this BIOS. From context, I think that mp260767 was referring to these two settings and implying that you should set them both the same (I can't think of any reason to set them different to each other, but maybe there is one).

They're in the BIOS as 'CPU - NB link speed' and 'NB - SB link speed'. Again, the settings are wrong and misleading. They are multipliers and not the stated speeds. The settings are actually 1x, 2x, 3x, 4x and 5x, set relative to the base speed. So if you increase the base speed, you must decrease these two settings in order to keep the two hypertransport links within spec (1GHz). To continue the 250MHz base speed example, you would need to set these two links to 4x, to get 4x250=1000. So you'd set them to "800MHz", which will give you 1GHz. Wrong and misleading labels abound in this BIOS.

The common suggestion, which I think is pointless, is to set the multipliers and dividers extremely low (to make sure that the CPU, memory and hypertransport links are running well under their rated speeds) and then see how far you can increase the base speed as far as you can. Since increasing the base speed has no effect at all on performance, I don't see why people do this other than as a competitive sport, to see who can get the highest base speed.
 
Angilion said:
Here's a really brief guide:

The common suggestion, which I think is pointless, is to set the multipliers and dividers extremely low (to make sure that the CPU, memory and hypertransport links are running well under their rated speeds) and then see how far you can increase the base speed as far as you can. Since increasing the base speed has no effect at all on performance, I don't see why people do this other than as a competitive sport, to see who can get the highest base speed.
Click to expand...

excellent post, i confirm the labels in the bios are wrong and misleading (especially the memory speed) - this should save a lot of time for 1st time asrock clockers :)
___________________________________
Asrock official 1.4 Bios
Opteron 146 (2.0GHz standard, 10x multiplier.)
Advanced / CPU Configuration:
Overclock Mode: CPU,PCIE Async
CPU Frequency (MHz) 260 (note: start at 240 and work upwards. 1.4 bios stops at 275. Also note OCW Beta BIOSs always report CPU speed at default 1.8GHz on boot, but it *has* changed really!)
PCIE Frequence (MHz) 100
Boot failure Guard: Enabled
Spread Spectrum: Disabled
Cool'n'Quiet: Disabled
Multiplier/Voltage Change: Manual
Processor Multiplier: x10 2000 MHz
Processor Voltage: 1.45 (note: max for this CPU. without m/b modification)

Memory Clock Auto
Flexibility Option: disabled - (enabling this will effect the divider setting - Clock 200mhz and enabled is 1/1 - 166 mhz and enabled 1/15 - auto and disabled 1/12 for PC3200)
Burst Length: 4 beats

Memory timings - will vary with RAM leave at auto or start with 3.0 / 4 / 8 / 4 and lower if running OK:

MA Timing Auto
DRAM Idle Timer: Disabled
User Config Mode: Auto

Advanced / Chipset Settings

CPU - NB Link Speed: 600MHz (800 for small overclocks up to 2.5 - lower with high overclocks 3.0+)
CPU - NB Link Width: Auto
NB - SB Link Speed: 600MHz (800 for small overclocks set same as NBL speed)
NB - SB Width: Auto
HTTtoPCI Bridge Decode Scheme: Auto
DRAM Voltage: High
Gated Clock Function: Disabled
PCIE1 Turbo Mode: Normal
___________________________________

i've just clocked my opty 146 to 2.6GHZ (asrock 1.4 bios) as above and run 3DMark05 but as you say i get no improvement from unclocked 2.0Ghz (results=3015 with a 9800pro)
no change -Why is this?
 
Last edited:
05 doesnt really take cpu speed into consideration when bechmarking as its heavily graphics card based ,probably better to use 3dmark01 to see a real difference as that benchmark is for full system
 
Nice one Angilion

You helped clear up some problems I was having with these stupid BIOS settings on my sons machine. You'd think ASUS (parent company?) could do better than that!
 
Guys,

I have an A64 4000+ San Diego and a 2GB Corsair C2PT Twinx Kit.

What sort of CPU overclock would I be looking at with this CPU (on air) and RAM ...and this motherboard?

I'm new to overclocking A64's and dont really overclock much as a rule anyways?

Suggestions??
 
voltmod

You probably need to do the voltmod (Google it). Its easy if you get it right but I suggest you use conductive paint as I got in a bit of a pickle with solder! I am very experience with soldering irons and such but nearly Borked my sons Mobo and it took hours and not a little luck to sort out my cockup (solder dropped onto a nearby components teminals and thats small stuff)!

I currently get 2.3 Gig stable from a 3200 Venice on standard cooler and approx 1.53 volts. My SanDiego 3700 gets 2.5 Gig stable with standard air cooler on an A8n Sli Premium but it's easier to O/C the Asus.
 
just out of curiosity who here has the OCZ voltage booster? i just bought one off ebay for my mobo for about £15. Is the warrenty on it the same type of warrenty you get for RAM?
 
Well I finally managed to get my rig together. My 146 went to 270htt no problemo, but 280 was a bit too far. I'm currrently on the 1.2 bios, so I hope updating this to one of the OCW bioses might allow me to go higher. Unusual findings are as follows.

(1) 3DMark2k3 was slower at 270htt (2.7GHz) than at 250htt (2.5GHz). I set the htt multiplier down to 3 for 2.7GHz (as opposed to 4 for 2.5GHz).
(2) My relatively cheap mem sticks (see sig below) managed 225MHz (i.e. 270htt x 5/6 ratio) in dual channel at 1t. The other ram settings were at default, however.

I might get a chance to do some further testing later.

Ginga
 
GingaNinja said:
(1) 3DMark2k3 was slower at 270htt (2.7GHz) than at 250htt (2.5GHz). I set the htt multiplier down to 3 for 2.7GHz (as opposed to 4 for 2.5GHz).
Click to expand...

Basic maths really, 4 x 250 > 3 x 270 :)
 
Xeon said:
Flexibility Option: disabled - (enabling this will effect the divider setting - Clock 200mhz and enabled is 1/1 - 166 mhz and enabled 1/15 - auto and disabled 1/12 for PC3200)
Click to expand...

edit: oops..small mistake - with clock set at 200mhz + flexibility option disabled is 1/1
(Note: if you are running standard pc3200 memory with this setting the overclock will be limited to about the max speed of the memory ~cpu 2.40 )

Clock 200mhz + flexibilty enabled is 1/15 (2/3) which infact is slower than 166mhz + disabled 1/12 (6/7) << recommended for pc3200.
Be aware that in the auto setting the bios shows pc3200 memory as 166mhz but it does infact run it at the correct speed.

Confusing aint it :p also Bios1.4 does not appear to have an HTT multiplier option.
 
Last edited:
When i'm running this in windows and i'm copying between the drives, i get massive slowdowns, i've populated both SATA ports and the SATA2 one, is this because i'm running the drives in IDE mode?
 
Check the windows device manager it should report the sata drives as scsi devices.
Disable raid + ide in the bios to test but you may need ide for your cd/cdrw/dvd drives.
 
@ MeddlE, thanks for that insight ;). The point is I wasn't expecting the the htt x multiplier product to have such an impact, overcoming the effect of 200MHz difference in raw processor speed. I have to admit though, I'm relatively new to all this and have much to learn.

Ginga
 
You must log in or register to reply here.
Back
Top Bottom