Asus A8N-E HTF stuck at 5x?

[gbohn]

Hi;

I recently tried to overclock my A8N-E (with Nforce 4 Ultra chipset), and have come to suspect that the Hyper Transport Frequency multiplier is 'stuck' at 5X, regardless of what I set it to in the BIOS. Is there any way to tell for sure?

I noticed that I could only get the FSB up to about 235 MHz, even if I set the CPU multiplier down to 7x (from 11x normal) and run the memory with a large divider). Even then, booting into windows at that speed corrupts my SATA drives data and I have to restore them. (Everything works great at all stock speeds though...).

I put everything back to 'auto', except the HTF multiplier. Running Nvidia 'Ntune' always shows the HTF at 5X regardless of what I set in the BIOS. The Nvidia website doesn't list the A8N-E as being supported by Ntune, so I don't know if I can trust it.

Running the Ntune 'Performance' benchmarks for Memory, Disk, and Video performance show no difference with HTF set to 1X or 5X.

Is there a benchmark that definitely should track with actual HTT speed? ANyone else have this problem

Thanks;

 

[The Pentium Guy]

Use CPU-Z, it'll tell you all you need to about your processor

 

[gbohn]

Originally posted by: The Pentium Guy
Use CPU-Z, it'll tell you all you need to about your processor

I have CPU-Z 1.28.6, and as far as I can tell it only shows the 'non-multiplied' frequency in the 'HTT' section (although it does show the CPU multiplier and resultant CPU frequency).

I couldn't find anywhere that it shows either the 'Hyper Transport Frequency' multiplier or the multiplied result for the final HTT frequency.

Am I interpreting this wrong?

 

[rise]

sandra will

 

[aporetic]

I just got my A8N-E Nforce4 Ultra Friday, and I'm getting similar results. I managed to get the "CPU frequency" up to 245mhz, but when I try for 250 WinXP won't load. I get a blue screen that flashes by too quickly to read.

I've lowered the CPU multiplier down to 6, set the HTT frequency to 3x, raised the timings and latency for the memory, set the memory frequency at 266, unplugged the SATA drives, lowered the pci timing to 33.3 mhz, and I just can't get anything higher than 245. (My cpu is a Venice 3000.) It didn't occur to me that the HTT multiplier might be the culprit. I thought maybe the Nforce chipset might be the bottleneck, even though I've read about other motherboards going as high as 300 mhz. I looked in the bios tweaks and didn't find an option for increasing the voltage to the chipset.

Any insights or advice is surely appreciated.
Thanks

 

[rise]

well, to read you bsod go to device manager/advanced/start up and recovery and uncheck "auto restart" at least you'll see the error code.

it most likely is some sort of memory issue but it could be your voltages, your psu, practically anything.

 

[gbohn]

Originally posted by: rise4310
sandra will

I tried Sandra 2005.SR1. So far, it looks like changing the HTF multiplier in BIOS has no effect on the reported HTT speed, or the results from a few of the Sandra benchmarks I tried.

Using Ntune changes the Sandra reported HTT frequency, but I still haven't seen this result in any change in the benchmarks performance...

 

[rise]

its very minimal look here--

http://www.neoseeker.com/Articles/Hardware/Guides/athlon64oc/2.html

 

[aporetic]

I verified with Sandra that lowering the HTT multiplier settings in the bios would not lower the resulting FSB. It is just as gbohn described - it's like the HTT multiplier is fixed or "stuck" at 5x. The solution (for now, anyway) is to flash the bios with version 1001. I tried 1003 and 1004-003 beta, and neither would allow the HTT multiplier to change from 5x (even though it appeared to in the bios).

After reading some of the posts here in the MB forum, I was concerned that version 1001 would not work with my Venice 3000+, but it works perfectly as far as I can tell. CPU-Z recognizes it as the Venice core and indicates the SSE3 support.

And the great news is that with tweaking, I'm getting 2.63 ghz stable!!

Have Fun

 

[gbohn]

Well, it looks like we can't hold out much hope of getting this fixed. I tried to report this as a problem to ASUS, and the response I got was:

"For Overclocking support please check with www.overclockers.com or www.tomshardware.com"

I guess they don't even want to investigate if this is a BIOS problem... Doesn't exactly make me happy I picked an ASUS motherboard.

 

[aporetic]

gbohn:

As I replied in my previous post, bios version 1001 allows all the bios tweaks to function properly. I'm currently running my system at 292 X 9 resulting in 2.623 Ghz. To achieve that, I had to set the HTT multiplier at 3x and set my cheap memory at the DDR266 setting. The limiting factor seems to be the CPU, because at one point I had the board FSB all the way up to 335 Mhz with the HTT multiplier at 3x.

Are you concerned that bios version 1001 will not work with your San Diego chip? I seem to remember reading a post here (in another thread) that the A8N-E with a San Diego CPU will at least boot with bios version 1001, so at worst you could at least re-flash with 1004 if 1001 won't work. My A8N-E came with 1001, and it booted fine, so I didn't have any reservations about reverting back to 1001 after realizing that the later versions had a problem with the HTT multiplier. I'm very pleased with the Asus board now, and I'm pretty sure they will fix the later bios revisions at some point.

 

[gbohn]

Originally posted by: aporetic
gbohn:

... and I'm pretty sure they will fix the later bios revisions at some point.

It seems to me that to be fixed, they have to both realize there's a problem and then want to fix it (assuming it's fixable).

Since they blew off my report, I'm thinking we haven't even made it to the 'realizing there's a problem' part...

 

[gbohn]

Well, things are looking pretty grim for the home team. I eventially talked to the tech. support people about this. Basically they said:

1) They don't believe that this ever used to work differently (in a prior BIOS). They think
this must be a feature of the processor itself.

and

2) They don't really care if it did actually work in that version.

I still haven't tried going to back to BIOS 1001, so I could only speak about my 1003 and 1004 experiences.

They said as long as it works at stock frequencies, that's all they care about.

They said that even if this did work in a prior BIOS level, the best I could do was call them back and tell them so they could make a note (and that they still would most likely not try to fix this).

This despite the fact the motherboard box comes with a whole section extolling the virtues of their 'advanced overclocking features'. How advanced is it when basic overclocking features won't work?

Makes me feel really great about purchasing ASUS. I'll have to keep this in mind when I make future purchasing recommendations...

 

[amheck]

I'm picking up the A8N-E today. Is that what will probably come loaded on the MB? If not, is there a way to "go back" to a previous bios version?

 

[gbohn]

O.k., I just tried the old 1001 BIOS, and surprise surprise, as I expected the stuck HTF multiplier problem is gone. With 1003, 1004, and 1005 the best "CPU Frequency" I could get (with the CPU and memory underclocked) was about 220 MHz without corrupting the SATA drives.

With a quick test on 1001, I was able to get over 245 MHZ and no SATA corruption.

So, the ASUS tech. support person said the CPU was causing the problem (despite being underclocked for testing) -- Wrong.

He said the HTF multiplier wouldn't work any differently with an older BIOS-- Wrong.

He said most people with the 3700+ SanDiego couldn't get the CPU frequency over about 220-225. Hmm... just coincidence that this coincides with the limit I found when the HTT frequency was being forceably overclocked? I think not...

With their entire reluctance to address the issue, I guess now I'm wondering if they just don't want this to work for people...

 

[amheck]

Well, that's definitely a bummer of sorts. Please let us know if you're missing anything major by going back to the old bios. I'm setting mine up today

 

[gbohn]

Originally posted by: amheck
Well, that's definitely a bummer of sorts. Please let us know if you're missing anything major by going back to the old bios. I'm setting mine up today

So far, the only bad thing (with very limited testing) is that there appears to be no 1T/2T memory options in the BIOS.

I called their tech. support again today (if only to report that I was right and they were wrong ). I explained that this multiplier used to work in 1001, but has been broken in 1003, 1004, and 1005.

Today's (paraphrased story) is that if the BIOS engineers changed it not work anymore, then that's the way it is. I asked well couldn't this be a problem that was unintentionally introduced? And what happens if someone finds a problem in the BIOS.

He said they have no mechanism to report BIOS problems, so what you get is what you get and there's no way of telling if they intended it to work this way or not. (I guess the HTF multiplier setting is now there just for ornamentation purposes...)

I pointed out that they were still updating the BIOS, and he said they (which I should have pointed out is supposed to be him in a way) only rely on their own testing. I asked if this meant that they must be infallible since they don't want to hear about potential problem... (At this point I imagined someone in BIOS support sitting there with their hands over their ears saying "La,La,La... I'm not listening..." )

So, I give up. I have to concede. Their ignorance appears to be invincible.

 

[aporetic]

The setting for the 1T/2T memory timing in bios 1001 is: Advanced/CPU Configuration/Dram Configuration. It's the second option.

I'm curious as to whether or not the Newcastle or Winchester core AMD 64 chips have the problem with the "fixed" HTT multiplier in the later bios versions (1003, 1004). Anyone??

It seems unfortunate that ASUS tech support seems uncooperative or unwilling to investigate the bios problems, and I can certainly understand gbohn's frustration. However, I am still optimistic that the problems will be resolved at some point. I don't consider myself an ASUS "fanboy" per se, but I have always been very satisfied with their motherboards over a period of 12 years, and the A8N-E is no exception. So far it has been rock-solid stable, has a good layout, and seems to be the same great quality that I have come to expect from ASUS.

 

[gbohn]

Originally posted by: aporetic
The setting for the 1T/2T memory timing in bios 1001 is: Advanced/CPU Configuration/Dram Configuration. It's the second option.

...

However, I am still optimistic that the problems will be resolved at some point.

On My system, with 1001, the first four options under Advanced/CPU/DRAM are:

Max Memclock
CAS #
RAS# to CAS # Delay
MIN RAS # Activation time

I still don't see any 1T/2T setting. Perhaps it doesn't recognize my memory in that version and I get a different set of selections?

I find it hard to be optimistic when all the people I've spoken with (at least 5 different individuals between email and phone) have basically told me they don't know if it's broken, don't want to know if it's broken, don't care if it's broken, and have no way to get it fixed even if they did care (which by the way, have we mentioned that we don't)...

To see them with the attitude "We have your money, so it just sucks to be you..." doesn't leave me with a warm, fuzzy feeling for the company despite how good or bad the hardware is.

And, if they don't want to supoport overclocking in the first place, how come the sales blurb for the A8N-E (See http://usa.asus.com/products/mb/socket939/a8n-e/overview.htm ) mentions:

" Precision Tweaker
Precision Tweaker is designed for serious overclockers. It offers ways to raise system performance inch-by-inch and step-by-step to achieve maximum performance! This is about getting the most out of your machine, and taking pride in your customized computer. Precision Tweaker provides overclocking options for all major parts of the system: CPU, Memory, PCI-Express, and Front Side Bus."

Oh, I'm not bitter ....

 

[blckgrffn]

Great.... sigh. I thought I just had a bum venice chip, I will look into this shenanigans tonight. Not to mention the damn thing won't boot at CAS2 with my ballistix at any speed :|

 

[gbohn]

O.k, after playing with this for a few days, I have a list of the 'bad' things I've found when running with the 1001 BIOS (at least with my setup).

1) As previously mentioned, I can't find 1T/2T Memory command rate setting
in the 1001 version (at least with my memory).

2) The HTF multiplier seems to go to 3x (as reported by Ntune) when set to
4X. I tried 2x, 3x, 4x, and 5X and only setting 4X seems to be off.

3) For some reason, going into standby (Win 2000) and then starting up again seeems
to reset my 'CPU speed' from 234 MHz to less that 200 MHz (effectively undoing any
overclock that was set). Mine shows something around 193 MHz in CPU-Z, which is
interesting since you couldn't set this (less than 200) on purpose as far as I can tell.

Rebooting recovers from this.

I don't know if this is better with the later BIOSs, because I didn't run overclocked
with those because of the stuck HTF multiplier issue.

So, at least with my setup, there are trade-off's to be made if you want to run with the 1001 level.

 

[amheck]

Sort of OT, but can anyone tell me where the RAM DIvider setting is in BIOS?

I'm trying to do this setting:
RAM divider = "100mhz 1:2"

Aaron

 

[imported_Technomancer]

Doesn't bode well for my X2 at all...

 

[gbohn]

Originally posted by: amheck
Sort of OT, but can anyone tell me where the RAM DIvider setting is in BIOS?

I'm trying to do this setting:
RAM divider = "100mhz 1:2"

Aaron

At least as far as I can tell, setting the memory 'Max. Memclock' is really a divider setting. So, setting this to 400 MHz gives 1:1, 333 gives 1:1.2, 200 MHz gives 1:2, etc.

It looks to me like the memory speed listed in BIOS/DRAM setting is the speed when the 'CPU Frequency' is at it's default 200 MHz setting. So I think setting '333' Mhz for the memory speed yields about the 'normal' 400 MHz speed if the 'CPU Frequency' is set to 240.

 

[ProviaFan]

Originally posted by: gbohn
At least as far as I can tell, setting the memory 'Max. Memclock' is really a divider setting. So, setting this to 400 MHz gives 1:1, 333 gives 1:1.2, 200 MHz gives 1:2, etc.

It looks to me like the memory speed listed in BIOS/DRAM setting is the speed when the 'CPU Frequency' is at it's default 200 MHz setting. So I think setting '333' Mhz for the memory speed yields about the 'normal' 400 MHz speed if the 'CPU Frequency' is set to 240.

Consider it confirmed. :thumbsup:

FWIW, I set HT at 4x in the BIOS, and when I booted into Windows, NV Monitor showed HT at 5x. I set it to 4x in the nTune application, and it seems to have changed (according to NVMonitor), but supposedly these settings are not "sticky." So, is it safe to run the HT at 1140 (I would think not), or do I just have to wait for another BIOS update to (hopefully) fix this before getting serious about the overclocking?