I think calling it 'NVIDIA Hypertransport' is a bit erroneous. It's an AMD creation, with little to no input from Nv, it's just been licensed to them.

(/nitpick)

Thanks Ryu. Now I know why I'm getting the error.

[q]I don't see your link Ryu can you give that again?[/q]

http://bbs.pcstats.com/viahardware/messageview.cfm?catid=19&threadi...

one is not holding one's breath that via will have the problems fixed on their next chipset.

I don't see your link Ryu can you give that again?

[q]I checked the event log and it is the VIA patch that failed to start.[/q]

Check out the link I submitted earlier. You'll find that pretty common behavior.

[q]I am considering getting an AMD 760 motherboard, but every motherboard seems to have a VIA 686b south bridge. Is this affected witht the same PCI problems? Or is it northbridge related?[/q]

The 760 + 686B combo is a "legacy" chipset design. The AMD 760 Northbridge still controls the PCI bus and connects to the 686B using that bus.

Newer chipset designs house the PCI bus in the Southbridge and use a custom high speed connection to connect the two core logic chipsets together.

Intel Hub: 266MB/s
VIA VLink: 266MB/s
NVIDIA Hypertransport (developed by AMD): 800MB/s

The AMD 760MPX chipset also moves away from the 133MB/s limit, but they use the 64bit 66MHz PCI bus as the connection point to the 768, which houses the 32bit 33MHz PCI slots.

http://www.amd.com/us-en/Processors/ProductInformation/0,,30_118_12...

In short the 760+686B combo does not suffer from the VIA PCI bug.

Here are the specification sheets for the 761:
http://www.amd.com/us-en/assets/content_type/white_papers_and_tech_...
http://www.amd.com/us-en/assets/content_type/white_papers_and_tech_...

[q]ag 25, PCI problems are the realm of the northbridg/q]

Only on older designs.

ag 25, PCI problems are the realm of the northbridge

I am considering getting an AMD 760 motherboard, but every motherboard seems to have a VIA 686b south bridge. Is this affected witht the same PCI problems? Or is it northbridge related?

What service or device driver, Gerbil #17? Run Event Viewer and find out -- that'll make the question real.

I checked the event log and it is the VIA patch that failed to start.

Craig P.

um, 360KB/s is higher than 1x CDrom speed

1x cdrom == 150KB/s 'ish

I'm using a computer with AMD 760 chipset inside. ZERO problems. It will probably last longer than my old Pentium Pro. Don't fall for the 10% higher Sandra mem score! Look at the bigger picture...

http://bbs.pcstats.com/viahardware/messageview.cfm?catid=19&threadi...

[q][i]
GeorgeBreese

Scripting Guru Date Posted: Feb/01/2002 9:20 PM

I am out of time, the relatives arrive tomorrow and the baby comes soon after that, but here's the technical scoop on the patch.

The patch contains all prior patches, except for the Promise IDE performance patch. The Promise patch has been replaced with the new RAID IDE performance patch.

The new RAID IDE patch scans the PCI bus for several varieties of "mass storage controller", such as IDE or RAID or SCSI. It skips VIA IDE controllers. If no appropriate controllers are found, it skips this patch and moves on to the other patches (MWQ, SBLive, and so on).

For every mass storage controller found, it sets the latency of that controller to 240 (0D=F0). It then sets the arbitration timer to 224CLKs (75=?7). It also sets a VLink performance timer (4B=8?), even if VLink is not present in the chipset.

Given my previous work on one particular KT266 board, I think this patch could very well cause SBLive crackling, unless the VLink timer change is preventing it. I won't have much (any?) time to test it, either.[/q][/i]

[q]I thought only a respin of silicon could fix this problem... well if this patch really works and eliminates the problem, good news indeed... I guess time and testing will tell...[/q]

The arbiter is still broken. This doesn't fix all the issues, it just increases bandwidth with certain cards. VIA is targeting the aspect of the bug everyone understands, low throughput, and ignoring the larger problem of bus sharing and low bandwidth sustained access.

The KT333 has an 8235 southbridge. Chances are it will be the first model with a corrected PCI bus design.

[i]What[/i] service or device driver, Gerbil #17? Run Event Viewer and find out -- that'll make the question real.

I install it under Windows 2000 and after booting I get a service or device driver failed. Anyone else had this?

keep the bastards honest: boycott VIA hardware.

Just a note, you have to reinstall GameTheatre XP drivers in XP after installing the patch. :)

Link does not seem to work. Anyone have another?

Originally Posted by dim-
Just one more reason for me not to buy VIA hardware. It\'s either AMD 760*
SiS 735/745
Or last but not least Intel.

OT: 'CAMP X-RAY' GUARDS: DETAINEES IN LEWD ACTS:
http://www.weeklystandard.com/Content/Public/Articles/000/000/000/8...

Actually, 150KB/s is generally considered 1x. In Nero, for example, burn speeds are all multiples of 150KB/s; 20X is listed as 3000KB/s, and 3000/20 = 150. I'm pretty sure my 2X CD-ROM claimed 300KB/s, waaaay back when; I'm *very* sure it didn't say 346KB/s :p

Nice 'bonus' until they actually decide to FIX THE DAMN HARDWARE isn't it?

Craig Im afraid you got your figures wrong there:

1X CDROM = 44100 (samples/second) X 2 bytes/sample X 2 channels = 44100 X 4 = 176400 bytes per second or ~173KB/sec

40X CDROM = appox 6.9MB / sec

This only makes you more right that the CDROM drive is not a real strain on the PCI bus ;)

George

AG#6 - The speeds of transfer (in MB/s) of a CD-rom, even on SCSI, are nowhere near enough to be affected. So you don't really need the patch, and if your system has no problems right now, why bother? See below for more explanation.

AG#4 - I doubt it, since they could never get these patches WHQL certified if they changed the PCI bus to run out of spec.

This patch is the same as the previous patch that VIA released, which tinkered with PCI bus priorities to give preferential treatment to Promise cards, but the list of cards has expanded to include Adaptec, Highpoint, and some others. This patch won't fix any sound card issues, they haven't been included. Besides, VIA can't give high priority to all the devices, for ovbious reasons. George's patch does a lot more, but it changes the PCI bus to run out of spec.

If you have a RAID array, probably the best option is the VIA patch linked up in the news article. If you have a sound card that crackels, that patch won't help, and you should get the latency patch written by Goerge (link at bottom). It looks from the TecChannel tests like VIA's raid patch gives better performance than George's latency patch, so that is the best for raid.

EO

http://download.viahardware.com/vlatency_v019.zip

1x CD-ROM = 360 kB per second.

40x CD-ROM = 14.1 MB per second

No, I don't think your CD-ROM would be affected, it won't come particularly close to the limits the hard drives were running into.

If I'm not using RAID, but am using an Adaptec SCSI card for my CDRom, will I see any benefit from this (or future) patches?

I hope this is a fix but only time and testing will tell.

They are probably just increasing the latency timers, which is illegal.

Originally Posted by GiGNiC
nice thoughts there #2

would this maybe finally get rid of the creative SB + VIA-southbridge issues ? for good that is ?