AVS › AVS Forum › Video Components › Home Theater Computers › My "exciting" ATI/865PEchipset/Hyperthreading/Tira story
New Posts  All Forums:Forum Nav:

My "exciting" ATI/865PEchipset/Hyperthreading/Tira story  

post #1 of 102
7/5/03 at 2:29pm
Thread Starter 
Hi guys,

this is quite a long story about my troubles with my new PC so I will give you a short introduction in case you do not want to read through all of this.

So far I have to once again bitch against ATI for having lousy, unstable drivers and not the greatest techsupport in the world causing me more trouble than necessary. Everyone here is saying that ATôs drivers have improved lately but they are still far from Nvidia´s stabilitywise. On the other hand I have to admit that ATI cards seem to still be better in conjunction with projectors using non standard resolutions like 1280x720.

Anyway, here´s the short summing up:

Since I assembled my latest machine consisting of

P4 2,6C
Intel 875/865PE chipset board (Asus P4P800, Gigabyte 8IK1100)
Ati Radeon card ( 9000, 9500 )
2x Twinmos Winbond 512MB PC 3200 MB DDR
TIRA Infrared USB adapter
Maudio 1010LT audiocard
Pinnacle PCTV Pro
onboard 3Com LAN
Antec 480 W True Power supply

... one month ago, I have been getting complete system freezes during playback of any video, watching TV or while running 3D Mark 03. I mean system completely freezes up, mouse does not move, latest sound is played over and over, no entry in Windows eventlog. Reset necessary.

After I tested and exchanged each and every hardware involved, tweaking any BIOS setting available and reinstalling Windows at least 20 times with different configurations, drivers etc. I have now come to the conclusion that the ATI cards or their drivers cause those freezes on systems like mine. How do I know ? Well, because the freezes completely disappear if an NVIDIA card is used. Unfortunately, it took me almost one month to realize the obvious.

Here´s the longs story for those who are interested:

Soon after the freezes appeared I found out that Tira was somehow involved in all this. If I had Tira connected and the plugin installed and loaded in Girder the system would completely freeze after starting a video or 3D application.

I then found out that those immediate freezes only occured if TIRA was connected to one of the USB controllers sharing IRQ with the AGP port. IF I used any other USB ports the freezes would still happen but much later (sometimes as late as 10 hours later while stressing the system using Prime 95, Burn in test and videoplayback, but the system would nevertheless always freeze during a 24 hour period )

I wrote Home Electronics ( Tira makers ) tech support and although very nice and helpful they could not help me much. They basically told me they never heared of Tira causing freezes ...

I then kept on searching ... I changed the mainboard, the RAM; the power supply, I added an USV, removed all cards, cables, and everything else I could think of, including the video card, but at that point I did not try an NVIDIA card yet, I exchanged my Radeon 9000 against a Radeon 9500. Of course they used the same drivers ... :mad:

Anyway ... the freezes remained. I then tried Windows 2000 with SR3 and later SR4 instead of Windows XP SR1, I tried each and every patch that came out after SR1, I tried diferent soundcards, Nics, soundcard drivers, etc. etc.

Nothing helped. I then started trying out different Catalyst drivers. 3.4, 3.2, later 3.5 and ... 2.5 ... well, that was the first succes. The freezes during video playback did not occur with Catalyst 2.5 but 3D Mark still crashed after a while ... so I kept on searching. I then found out that deactivating Hyperthreading in the BIOS completely solved the problems.

I left 3D Mark 03 running for 30 hours while running Prim95 and Sandra Burn in test ... completely stable.

So I wrote ATI tech support to tell them that their drivers were not compatible with Hyperthreading. They insisted that they were and that there were "no know issues with Hyperthreading". Instead they implied I look elsewhere ( Tira ) for help after hearing that Tira on the USB controller sharing IRQ with the AGP port would exaggerate the freezes.

Well, what was I to do ? I sort of believed them so I wrote Home Electronics again asking for a refund. But somehow, after a few days of living without Hyperthreading I got mad. This could not be it ! Why do I have to turn off something that supposedly gives me better system response and which I paid for just to get a stable system ?

So I then started another round of tests, inquiries ( I wrote Asus and Intel tech support etc. ) and after a few days I thought about getting an NVIDIA card to test.

What can I tell. Installed the card and drivers and started the all the previous "bad" things: Tira on USB controller sharing IRQ with AGP port, starting video, running 2 instances of Prime 95, Burn in test, Sandra burn in test and 3D Mark ...

has been running stable for 34 hours. I guess this proves that ATI drivers are indeed the problem here. Yesterday I wrote them again telling them exactly what I wrote here and suddenly, 1 hour later I got a response informing me about a hotfix for Radeon 9600 cards and Intel 875 chipset problems. Unfortunately, I could not install this hotfix, maybe because my card is no Radeon 9600.

Anyway ... this whole story has kept me busy for the last month or so ... and once again I have to conclude that ATI has a LONG WAY ahead until their drivers matches NVIDIA´s stabilitywise.

Also I am sick and tired of hearing this "there is no know compatability problem" response when writing tech support. If it were known, I probably would not write them in the first place so why do they keep telling me this ?

Unfortunately the NVIDIA card I tried ( Asus 9560 FX ) did not work at all with my Sim2 HT300+ projector. I could not get a picture @ 1280x720 even after playing with the latest Powerstrip beta so I returned it today. What a pity. Once card has great driver support and bad HTPC compatability, the other one is the opposite. :mad:

I really hope that ATI will fix this problem soon. I know it may not be a very common problem ( haven´t found anyone else in any forum yet that has the same problems ) but it is definitely still a major problem.

Best wishes,
Alex

AVS Top Picks

post #2 of 102
7/5/03 at 4:12pm
A lot of people are having memory problems with the 875/865 chipsets....future bios revisions for the motherboards may help.
post #3 of 102
7/5/03 at 4:39pm
Do you need to disable the on board audio if yo have an additional audio card? Would check asusboards on bios etc. I am not sure about Tira or the pctv board but I had problems with HiPix and usb 2.0 enabled. I am missing why ATI is at fault as it works fine with Hyperthreading. I dont use it now because of MyHD playback does not work with it (but at least it works with usb 2).

Tim
post #4 of 102
7/5/03 at 6:57pm
Tim,

Are you saying you can't play back MyHD files if hyperthreading is enabled? Which mobo are you using? What memory? This is a heads up for me because I'm just about to put together a system like this.

Thanks,
Paul
post #5 of 102
7/5/03 at 7:56pm
Alex,

I use the P4P800 board with hyperthreading disabled. Everything is fine.

I think that some cards still have issues with hyperthreading. For example, the problem could even be your m-audio card. Many drivers were developed pre-hyperthreading.

I would not assume that the ATI drivers are necessarily at fault.

Maybe you should try Windows XP. It was developed to supprt hyperthreading as well. I run WinXP+SP1+DX9A and it's been rock solid in general.

Good luck!
post #6 of 102
7/5/03 at 8:04pm
Quote:
Originally posted by automagic
Tim,

Are you saying you can't play back MyHD files if hyperthreading is enabled? Which mobo are you using? What memory? This is a heads up for me because I'm just about to put together a system like this.

Thanks,
Paul
Paul,

That is correct. It is a known issue with the MyHD Card and is being worked.

Tim
post #7 of 102
7/5/03 at 8:34pm
I run an Asus P4P800 with P4 3.0C, HT enabled, and ATI 9800Pro on my new game machine. Running extremely stable, no lockups here.
post #8 of 102
7/5/03 at 8:51pm
Does "P4 2,6C" mean a 2.6GHz Pentium 4?

I keep seeing things saying that Hyperthreading only works on P4s faster than 3.06GHz...

http://www.intel.com/design/chipsets...rd/valuepc.htm
"Hyper-Threading Technology requires a computer system with an Intel® Pentium® 4 processor at 3.06 GHz or higher, a chipset and BIOS that utilize this technology, and an operating system that includes optimization for this technology. Performance will vary depending on the specific hardware and software you use. See www.intel.com/info/hyperthreading for information."
post #9 of 102
7/5/03 at 8:56pm
I guess maybe Intel didn't update their site, as it looks like they have 2.4 and 2.6GHz models that support Hyperthreading now.

===========================

Apparently ATI is releasing a motherboard chipset (RS300) that provides Hyperthreading support and integrated Radeon 9200 functions...

http://www.neoseeker.com/news/articl...Hardware/2506/
post #10 of 102
7/5/03 at 9:04pm
2.6C

means a 2.6ghz pentium 4 processor on a 800mhz bus

b=533mhz processor side bus
c=800mhz
post #11 of 102
7/5/03 at 10:43pm
Quote:
That is correct. It is a known issue with the MyHD Card and is being worked.
Generally devices that have problems with hyperthreading also have problems with multiprocessor systems that don't feature hyperthreading. So, it's more of a problem of having multiple processors executing different threads at the same time regardless of whether those processors are actual physical processors or "logical " processors as is the case with hyperthreading. The HiDTV card has this problem as well. I have one in my dual Xeon machine and the only way I can get the card to work reliablely is by disabling one of my processors. Other people with single CPU hyperthreaded systems have had the same problems with the HiDTV card. Aren't both of those cards (MyHD and HiDTV) Janis based cards? If so then this may be a problem with all the Janis based HDTV tuner cards. Anyway, I decided to try a Fusion II card in my next system, which I will be building fairly soon. I hope to have better luck with it.

Anyway, the thing to do, is stay away from the hardware devices that have these problems and instead buy hardware from companies that know how to make their drivers thread-safe. In this day and age there is no excuse for these hardware vendors not to write thread safe device drivers for their products and if they refuse to do so they need to get out of the business. In the past they could simply use the excuse that multiprocessor systems were not mainstream and therefore not supported by their products. However, since Intel is now including Hyperthreading support in all their newer desktop processors and has made it very clear that they would do so for some time now, that excuse is no longer valid and the companies that cannot adapt need to die.

Another thing that needs to be done is exactly what you've done here and that is to post threads like this informing us all of these problem products that don't have thread safe drivers. Thank you very much. I had actually thought about the possibilty of going with an ATI product in my next system, but now I know I better stay away from those products as they will likely cause problems for me in a dual Xeon system.

Quote:
Maybe you should try Windows XP. It was developed to supprt hyperthreading as well.
You can try XP if you want, but I doubt that will solve these kinds of issues. The only major difference with regards to XP and the way it treats hyperthreading is that it recognises the difference between logical and physical processors when considering the number of processors allowed under the license. For example, if you have a dual Xeon system with hyperthreading, XP proffesional will allow you to have all four logical processors even through the license only allows 2 processors because it realises that the 4 processors are logical processors and that there are only 2 physical processors present. Windows 2000 doesn't recognise that difference and therefore youd need windows 2000 server or higher in order to have all 4 logical processors in use since windows 2000 proffessional only allows 2.
post #12 of 102
7/5/03 at 10:55pm
Quote:
Originally posted by automagic
Tim,

Are you saying you can't play back MyHD files if hyperthreading is enabled? Which mobo are you using? What memory? This is a heads up for me because I'm just about to put together a system like this.

Thanks,
Paul
Stay away form the HiDTV card also as it too has the same problem with either Hyperthreaded or multiprocessors systems.
post #13 of 102
7/6/03 at 1:38am
Thread Starter 
Hi !

Guys, please .... after one month of testing EVERYTHING POSSIBLE, there is no chance in the world, that something else except the ATI card is causing the freezes.

---------------------------------------------------------------------------------------------
edit: Of course it may be that Asus could resolve that issue with a BIOS update but you see ... I also tried a Gigabyte mainboard with 875 PE chipset and had exactly the same problems, so it is either Intel who must revise their chipset drivers or ATI.
----------------------------------------------------------------------------------------------

Ron Ricko: It is not a memory problem

rudolpht: Of course I have the latest Asus Bios but the point is: everything works well with NVIDIA cards. Since I checked everything else, thats the only piece of hardware that made the difference between stable and non stable.

Rich4AV: No, the Maudio card is not the problem because I took it out any tried it with the onboard soundcard. Still results in freezes.

Olyar15: As I said in my first message, the problem sometimes only occurs if you run the machine for 10 hours or more under heavy load. Tira was the only device that accelerated those freezes so they happened instantly when an ATI card was installed. ( not so with NVIDIA cards )

Now, I know from other forums that a lot of people run ATI cards on 875/865 chuipset boards without any problem. ATI´s website that hosts the hotfix for the Radeon 9600 says that those problems only occur on some systems with many partitions, large harddiscs etc. It may well be that your system does not have the problem but that does not mean that there is no problem.

Those freezes are totally reproducable. They appear instantly if Tira is installed on the USB ports sharing IRQ with the AGP port or during a 24 hour stress test while video is playing if Tira is not connected or on other USB ports.

Catalyst driver version 2.5 cures this problem but it happens with all 3.x drivers and 3D Mark will still crash after a while.
Turning Hyperthreading off also cures the problem and also allows for 3DMark to run without crashing.

Maybe someone here has the ability to contact someone @ ATI tech support personally. I did write to them but it would increase pressure if more people contacted them of course.

Thanks,
midiboy
post #14 of 102
7/6/03 at 2:18am
Midiboy,

I am greatly indebted to you for bringing this up, but the moral of this story seems to be, we turn off hyperthreading and we all live happily ever after.

Paul
post #15 of 102
7/6/03 at 2:45am
Have you done a "clean" install of the ATI drivers? If not, you might consider using DRIVERCLEANER which (along with instructions) is available on http://www.driverheaven.net (DriverHeaven).

If the problem is related to an IRQ being shared by the USB and AGP, would it not be best to re-arrange things? Also, have you checked the latency for each?

And, of course, is(are) the BIOS(es) the absolute "latest and greatest"?
post #16 of 102
7/6/03 at 3:07am
Thread Starter 
HI !

Keh: Yes, of course I have used Driver Cleaner, not only that, I had build at least 20 complete fresh Windows installations during this month to rule out any driver conflict issues.

How can I rearrange the USB sharing IRQ with AGP port issue ? This is a fixed setting on the mobo. Two of the 4 USB controllers that exist on this board share IRQ with the AGP port. As I wrote earlier, I did try TIRA in one of the ports NOT sharing IRQ with the AGP port but this only slowed the problem. System did not freeze immediately when opening a video but a few minutes/hours later.

Of course I rearranged the PCI cards I have ( only 2: audio and TV card ) but those cards did not affect the freezing issues anyway so ....

With latency do you mean PCI latency ? To my knowledge this does not affect the USB controllers since they do not run off of the PCI bus.

Yes, the BIOS is the absolute latest and ( possibly :D ) greatest !

@automagic: No, thats not what I intend to do ! Personally I think Hyperthreading is a great concept even if there is no proof yet that it really helps with speed. But I can tell you this: if Hyperthreading is on and my system is stressed out completely ( 2x Prime 95, Burn in test, Sandra Burn in test ) and I start audio playback, everything is smooth. If I do the same without Hyperthreading I will get clicks and pops etc ... The same goes for videoplayback. It is much smoother if you do something next to it with Hyperthreading as opposed to disabling it.

So my wish is to keep it turned on. ATI or any other manufacturer should finally get over the issues involved with it ! It´s not THAT new a technology anymore. It´s been enabled on the XEON processors for quite some time now as far as I know ...

midiboy
post #17 of 102
7/6/03 at 3:25am
Thread Starter 
.. ah ... did not mention that yet ..

before someone comes up with the temps issue ( usually, this is the first reply I get on forums :D ) ...

No, it is not temp problem either. This was one of my first suspects. my processor temp is at 38 degree Celsius idle and max. 49/50 Celius under heavy sustained load. I even bought a Zalman CU7000 cooler an AS3 to cover that issue once and for all.

Regarding overclocking ... of course I am doing a little :-) But do not jump to any conclusions ... during the month of testing this, my system ran at stock speed. So no overclocking issues either.

I just mention this because I would really like to stress the point that I tried everything I could think of during the last month. Of course I did not try each and every video card out there but I tried a Sapphire Radeon 9000 and Radeon 9500 and an Asus V9560. IT freezes with the first two and not with the latter. To me thats proof enough.

I do agree though that it could be something IRQ related since Tira speeds up the freezes tremendously. Strangely, any other USB device I had did not have this effect. I tried an USB printer ( HP 930C ), a Bluetooth dongle (TDK) and some USB2seriell converters on the USB ports sharing IRQ with the AGP port. The only USB device affecting the freezes was Tira and only if the Tira plugin was loaded and active in Girder.

midiboy
post #18 of 102
7/6/03 at 8:29am
The reality is that Hyperthreading helps in rare examples, e.g., photoshop transforms. Don't get me wrong, I wanted it to work, I like the concept, and I was first on the block to by an original 3.0 GHz to get the feature (at a very high price premium).

It is too bad, maybe post Northwood processors will do better.

In the meantime, I would suggest a proven ATI model, e.g., 9700pro or 9800. The 9600 does have some DirectX issues as the newest model. I know there is a price premium, but there is also a time-value of effort. With high proportion of forum users and Mike Parker modding ATI boards, they have to be doing something right (but I have been using them since their original IBM 8514/a compatible adapter).

I assume the alternative would be an Nvidia or other card. That or try an excellent USB-UIRT2 tranceiver instead of the Tira.

Tim
post #19 of 102
7/6/03 at 9:18am
Thread Starter 
@rudolpht:

I know you´re trying to help but having tried old and "proved" ( as you put it ) technology like the Radeon 9000 and Radeon 9500 I cannot believe that the Radeon 9700 or 9800 would not cause the same freezes. I do not believe that the chips itself are the problem. Since using an older driver version helps it probably is a driver problem. Having said that, the newer cards will probably also freeze on my system.

The Radeon 9700 pro by the way is just a little bit quicker than the 9700 and the latter has currently the absolute best price/performance ratio. The 9800 is far too expensive compared to the 9700 but thats not the issue here. :p

As far as Tira is concerned, I have no idea if the UIRT will be better or worse in this regard but it does not matter. Once again, the freezes will also happen without Tira connected, only much later. Still, this is unacceptable for any kind of work, even it just involves watching a movie.

In my opinion ATI and Home Electronics should get together to solve this together since Tira influences the situation but the real fault seems to be with the ATI drivers.

midiboy
post #20 of 102
7/6/03 at 9:28am
This is just info. This won't help you, sry. ;)

I have dual processor issue with my RADEON 9000. It crashes once per one or two week(s) under 24/7 running condition. When I reboot machine after the crash, Windows XP SP1 tell me it is crashed in the middle of ATi driver. In my case, this is hard to reproduce and I don't need to use RADEON since this is not my media machine. Therefore, I'm using RAGE FURY PRO and it works perfect without any problem.

I want to know if any of other members have issue or no issue with RADEON on SMP/HT environment. :D
post #21 of 102
7/6/03 at 10:27am
Do all of the 800Mhz BUS CPUs ('C' models) support Hyperthreading?
post #22 of 102
7/6/03 at 3:47pm
Thread Starter 
@ Kazushi:

would you please tell ATI about those problems ? I think you should not simply accept the fact that your Radeon crashes your system every 2 weeks or so ... this is not acceptable in my opinion and you would do yourself, me and others ( including ATI ) a great deed if you could help them improve their driver stability.

At least I am not the only one with those crashes/freezes, even if our system may be different my problems my be related to yours ...

@ PVR: As far as I know yes !

midiboy
post #23 of 102
7/7/03 at 5:56am
midiboy:

OK, it's a clean install (sorry if the question upset you, but it had to be asked..suggestions are made not only in an attempt to assist the person asking, but anyone reading the post).

Yes, I meant the PCI latency. Generally speaking, the BIOS will allow you to set an 'overall' default value, but using something like POWERSTRIP, you can tweak the values for each device.

If your USB controller(s) are not on the PCI bus, on what bus are they?

On the subject of 're-arranging', there are two factors -

a)The 'interrupt line' used by the device to signal the BIOS. There are four of these, with each slot/on-board device given an order of preference. When there are more than four slots/on-board devices, then some will have to share the same 'interrupt line' as "first choice". A very comprehensive BIOS will allow you to modify its table in the advanced setup, but often this requires use of a BIOS editor, or simply cannot be done. The usual technique is to simply swap the card locations to get a different 'interrupt line' selection. Obviously, this cannot be done for on-board devices or the AGP port.

At startup, the BIOS 'negociates' with the devices to first come up with an 'acceptable' arrangement of 'interrupt lines' and then

b) negociates a 'logical flag' (IRQ) to be used to pass the interrupt notification to the OS.

But....

Depending on a setting in the BIOS, "Plug and Play Aware OS", Windows (which avoids going through the BIOS as much as possible) will happily ignore the results of the BIOS 'negotiation' and do its own version, and in one of various ways.

If the BIOS has been flagged "ACPI Aware OS", then instead of spreading the IRQs over the 16 available, it crams them down to as few as possible and uses its own internal mechanisms to determine who really is responsible for raising the interrupt.

But if the BIOS is flagged "IO APIC Support", then an advanced form of IRQ is used that provides 24 values. Switching this on or off generally requires a re-install of Windows.

(Your BIOS will likely use different 'labels' for these features.)

So when it comes to IRQs, it really is a question of who's doing what with whom and how...

One other point - "Random" problems are often power related. I note that when the Tira is connected the failure is immediate, when not, the failure is intermittant. How much power is the Tira pulling from the USB?

Yes, I noted it was a Truepower 480 and its tight regulation, but there have been a disconcerting number of posts (on another forum) about the fact that the 'center point' of the rail, particularly the +5.0, was low (confirmed by my own experience). You should monitor your voltages under load.

And it may not be the PSU per se, but the mobo's ability to pass power to the USB; a powered hub might be needed.
post #24 of 102
7/7/03 at 8:22am
Thread Starter 
Hi Keh,

wow ... you seem to know what you´re talking about. Well, I´ll try to answer those points you brought up although I still do not see why it seems that everyone is looking for the most bizarre explanation for the freezes instead of simply accepting the fact that ATI drivers are causing them. Remember, with an NVIDIA card the freezes completely disappear even with Tira connected so it is quite unlogical in my opinion that it could be a power or voltage problem.

Anway ... let´s finish this power issue while we´re at it: I do have a powererd USB hub and I tried it, but it did not change anything. I also tried an Enermax 365VE ... (don´t remember the full name anymore) power supply, and it also did not change anything. My 5V line never falls below 4,972 V even under full load with all USB devices connected. I am monitoring this and the temps using Asus PC Probe. Of course I do not know if those values are really accurate but I do not have a multimeter or so ...

The fact that Tira will only cause those immediate freezes during video playback if it is connected to one of the USB controllers sharing IRQ with the AGP port also implies that it is more of an IRQ sharing issue than a power or voltage related issue, don´t you think so too ?

Also, those immediate freezes only occure if the Tira plugin is loaded and active in Girder. If the plugin is not loaded ( but Tira is installed and connected ) it behaves "normal" ... meaning, the freezes will occur during video playback, not immediately but over the next few hours.


Quote:
If your USB controller(s) are not on the PCI bus, on what bus are they?
As far as I know the 4 internal USB controllers on the 865/875 chipset are part of the ICH5 southbridge or Hub as Intel calls it so they are not running of the PCI bus. The same applies to the SATA ports and the parallel ATA ports.

So adjusting PCI latency will not affect the USB bus at all. IT will only affect the PCI bus.

Regarding IRQ´s ... I am using APIC mode so I have 24 IRQ´s available in Windows XP. I have attached a complete system report I made with Sisoft Sandra in case you or anyone else is interested. I have also attached my IRQ settings at the end of this document. You can see that each of the 2 PCI cards I use ( PCTV Pro and 1010LT soundcard ) are in their own IRQ. You can also see the 2 USB controllers sharing IRQ with the Radeon 9000. This I cannot change or influence in any way.

Some of the entries on this system log are in German, because I live in Austria, but I think you should not have any trouble identifying things, most of them are English anyway.

Remember though ...

Running this system as it is with Radeon Catalyst 2.5 drivers and Hyperthreading ON will result in a stable system if I do not run 3D Mark 03 ( This is what I am running right now, it has run completely stable for the last 45 hourse running, 2 instances of Prime 95, Sandra burn in test, Burn in test pro 3.0, videoplayback and audioplayback in a loop in addition to normal websurfing )

Running this system with Hyperthreading off an any of the Catalyst 3.x drivers will also result in a completely stable system, now I can even run 3D mark 03.

Running this system with Hyperthreading on and any of the Catalyst 3.x drivers will result in freezes ( immediate or later on depending on Tira )

So to me this is definitely a simple ATI driver issue, no way how you put it, especially if you consider the fact that with NVIDIA cards ...

but I have said this numerous times before, right ?

midiboy

 

system summary.zip 8.9716796875k . file
post #25 of 102
7/7/03 at 9:32am
Except if it is ATI related but still more H/W than driver related. One reason I suggested the 9700 was the additional power supplied to the card itself, above and beyond the bus.

Tim
post #26 of 102
7/7/03 at 9:38am
Thread Starter 
Quote:
Except if it is ATI related but still more H/W than driver related. One reason I suggested the 9700 was the additional power supplied to the card itself, above and beyond the bus.
rudolpht: No, the Radeon 9500 is the same in this regard. One has to supply extra power to this card too. So once again, I cannot believe that the Radeon 9700 or 9800 will behave any different in my system. On the contrary, whith those cards I won´t be able to use the Catalyst 2.5 drivers since the new chips were not supported with the old driver version.

midiboy
post #27 of 102
7/7/03 at 9:44am
The fact that the problem exists with an ATI card and not with one from nVidia simply indicates that a conflict exists with the ATI scenario; it does not necessarily mean that the conflict is an ATI problem. Nor does it mean that it is not - all that can be said is that a conflict exists.

One would expect that if there were a driver problem, sites such as the aforementioned DriverHeaven or Rage3D would be rife with complaints.

I note that Omega's driver mod of the 3.5 is now available - since his are tweaked for stability and image quality, you might want to try it, but I doubt it will be a solution.

Please remember that you are on the "bleeding edge" (and are reading the ramblings of a lowly AMD1700+/Sis735/ATI7200 user).

On the power side, I would expect a Truepower to be dead accurate; that 25% overvoltage on the +12.0 made me raise an eyebrow, but that likely only makes your fans spin a bit faster. And that reading is at 'idle' (a multimeter is no better than the mobo's sensors - it cannot react to the wild variations of load that occur from one microsecond to the next).

Enabling APIC adds another 'new' factor into the mix; is it required for hyperthreading? I note that although IRQs 5, 7, 11, 17 are available, the AGP port is shared with a USB controller. Did the same occur with nVidia? Have you excluded them (legacy) in the BIOS? Have you checked with M$ as to any problems/fixes in this area? I am loathe to suggest you turn off APIC, as this necessitates a re-install, but there is always that point of despiration.

A 'bus' is a shared pathway for data flow to/from the devices (be they stuck in card slots, chips soldered to the mobo or embedded in a multifunction 'superchip') requiring a common form of contention resolution/control, such as an IRQ. Controllers (USB, ATA etc) are 'devices'.
post #28 of 102
7/7/03 at 11:40am
Thread Starter 
@Keh:

I cannot use the NVIDIA card anymore, I returned it to the Shop already and to be honest, I cannot ask them to lend it to me a third time. In the course of one month I have borrowed and returned possibly 10 to 20 hardware parts and the last time they were starting to get a little bit upset. I don´´t want to push it.

Unfortunately, I did not "record" the scenario with an NVIDIA card present. I just left the machine running for some 34 hours to check for freezes and in the meantime I was busy trying to get it to work with my projector (which failed, I could not get the card to display any picture @ 1280x720 on my Sim2 HT300+ even trying the latest Powerstrip beta)

I have no idea if the IRQ´s changed with the NVIDIA card present in the system but I do not believe it. Anyway ... 2 USB controllers are always sharing IRQ with the AGP port, that was the same during each and every installation on this board. But since I also tried an Gigabyte 8IK1100 board, which certainly has a different IRQ sharing scheme and experienced the same freezes I once again think that this is somehow an ATI driver issue.

Are you working for ATI or why do you have such a hard time accepting the idea that it could be an ATI driver issue ? :rolleyes: :D

I have no idea why the 12v rail was at 15v while running Sandra. Right now it is @ 12.038V ....

I have read numerous times that APIC mode is necessary for new boards to be able to use every device on the board. Momentarily I am a little reluctant to try out yet another fresh installation. Had a little to much of this during the last month or so :rolleyes:

Quote:
I note that although IRQs 5, 7, 11, 17 are available, the AGP port is shared with a USB controller.
Yes, because thats how the board is hardwired as I understood it. But I may be wrong.

Quote:
Did the same occur with nVidia?
As I said: don´t know.

Quote:
Have you excluded them (legacy) in the BIOS?
No, I have not excluded anything. What should I exclude and why ? If it is indeed needed, why does it all work fine with driver V 2.5 ? Why does it work without Hyperthreading ?

Quote:
Have you checked with M$ as to any problems/fixes in this area? I
I did not write them specifically but I installed each and every patch that came out after SR1. Approx. 55 patches or so. This includes the new kernel update that supposedly fixes a HT related issue.

Quote:
I am loathe to suggest you turn off APIC, as this necessitates a re-install, but there is always that point of despiration.
As I said, I am a bit reluctant to try this but I may do so if ATI wants me to. In any case, the next steps have to come from ATI in my opinion. I wrote them on Friday that I could not install the radeon 9600 hotfix on my system and am waiting for a reply to this since. I hope they will do something about this soon.

I wanted to buy a RAdeon 9700 some time ago already but I will wait with this until my problems are solved because with a Radeon 9700 I would not be able to use the Catalyst 2.5 driver anymore.

Ah yes ... Omega drivers. Heared of them. I might try them soon.

Quote:
One would expect that if there were a driver problem, sites such as the aforementioned DriverHeaven or Rage3D would be rife with complaints.
Yes, I know and I am very sad that I seem to be the only one who has those problems. Or maybe the others did not stress test their machines yet or simply have a slightly different configuration.

It would be easier if the problem were more widespread, at least I would be taken more seriously :D

Ciao,
midiboy
post #29 of 102
7/7/03 at 12:35pm
Midiboy,

I've sent my problem reports to ATi.
post #30 of 102
7/7/03 at 1:21pm
I think that if you reread the first two paragraphs of my last response, you will note that I am neutral on the 'cause' of the problem, and the reason for that viewpoint (which you acknowledge at the end of your post). I would suggest that a reread of your own posts would tend to leave an impression of a 'fixation' on one possibility to the exclusion of any other.

If you are convinced that it is a driver problem, you would be better served to take your crusade to the 'gamer' sites - they have a far greater impact than the HTPC crowd which seems to get only an annual perusal from ATI.

And this absolves you from any further requirement for investigation - you need only bemoan your fate and await a 'fix'.

Your BIOS setup likely has the provision to declare most IRQs to be "plug and play" (which means they can be assigned to a device through 'negociation' by the BIOS or Windows) or to be 'legacy' (which means no assignment is negociated, as the IRQ to be used is 'hardwired' to the device (either by jumpers, or a parameter in the driver startup) and it is your responsibility to ensure that no two devices both try to use that IRQ if they cannot 'share').

The common 'legacy' usage of IRQ5 was the DOS Soundblaster (which in turn was a 'legacy' of an earlier sound card).

The common 'legacy' usage of IRQ7 was interrupt driven (as opposed to polled) parallel printer port or some other sound cards.

The common 'legacy' usage of IRQ3&4 was/is COM2/4&COM1/3.

All other IRQs below 9 are fixed to mobo operations.


But since you have Powerstrip, you might try looking at the PCI latency for each device.

For those unfamiliar with the term, 'PCI latency' is the length of time that the device 'owns' the bus, preventing other devices from using it. If the latency is higher than needed, then other devices are 'blocked' unnecessarily and the system is slowed (unless the other device is a 'real-time' function, in which case things can go very bad indeed). If the latency is too low, the data transfer can be interrupted before it is finished, resulting in an error and retry (if not corruption).

Since card manufacturers have no idea what level of mobo they will run under, they tend to default to extremely high latencies and ignore BIOS defaults.

I run with a BIOS default of 32, play it safe and run the 7200 at 64.
New Posts  All Forums:Forum Nav:
  Return Home
  Back to Forum: Home Theater Computers
This thread is locked  

AVS Top Picks

AVS › AVS Forum › Video Components › Home Theater Computers › My "exciting" ATI/865PEchipset/Hyperthreading/Tira story