We all knew VIA has had a bad rap, but Damn...

[Ryu Connor]

BUT 4.2G of memory bandwidth does not fit through a 2.1g bus, so the bandwidth is wasted unles you use the integrated video...which is not a candidate for me.

Err.

Did you even read my post?

The 4.2GB/s bandwidth can be used by the integrated video, the AGP slot, and is connected to the hypertransport pipe direct to the MCP. Meaning all the integrated peripherals and PCI bus are attached to it.

In other words 2.1GB/s for the processor and a spare 2.1GB/s for all your I/O devices.

[Ryu Connor]

A picture is worth a thousand words.

Twinbank in all it's glory.

[Marshmallowman]

form an earlier post by me

instead you have 3.2 GB/s of bandwidth that is only partially useable by the Integrated video and some busmastering DMA on PCI cards

WIHOUT integrated video..you would probably be lucky to use %60 of the available memory bandwidth
(3.2 should have read 4.2)

The real point I am trying to make is why did the not have it set up so you had a possibility of using two banks with slower ram, eg save $$$ because you can use two banks of cas2 pc1600(or 2100) instead of using expen$ive pc3200.

have you seen the twin bank serverworks chipset?, actual it may have been 4 banks of pc100 sdram..

and the intel chipset with two banks of pc1600, 2x200ddr->400qdr perfect match, highperfomance will relatively cheap cas2 pc1600 parts.

the nv chisets are looking good, but I think they missed adding a nice fat feature

[Ryu Connor]

The real point I am trying to make is why did the not have it set up so you had a possibility of using two banks with slower ram, eg save $$$ because you can use two banks of cas2 pc1600(or 2100) instead of using expen$ive pc3200.

I posted earlier telling you that you could. The board has all the necessary dividers to run memory at 100 or 133, etc.

[Greebe]

You mean the chip will support it... I doubt any one has seen a reference board yet, let alone an actual product.

[Ryu Connor]

Well if you want to get real technical the IGP/SPP supports it, but then the firmware has to have the necesarry field set to manipulate the register in order to specify a value.

Smart ass.

[Greebe]

Well the women like it

[KvHagedorn]

Doc, have you tried out the RT board on a 760MPX board yet? If it pushes things so much perhaps Matrox should make a 64bit PCI version..

[Greebe]

From what I'm gathering any and all 760MPX boards have hardware (chipset)/BIOS issues. AMD has some work to go before this would be a tangible product IMHO.

[KvHagedorn]

I hadn't heard of any serious issues before. Does anyone with an MP or MPX board have any experiences to share about this?

[MultimediaMan]

I can't even get my 760MPX (Chaintech 7KDD) Board to completely boot with a Parhelia. It can POST, but it hangs on the DMI pool screen: I've tried EVERY possible permutation (right down to a sheet of paper to make ticks mark about what I did and need to do next...) of the Video setups in the BIOS to get my P to work, to no avail.

Many others are experiencing issues as well. I dunno if this one is gonna get fixed anytime soon: there are a number of issues with PCI 64bit or 66MHz cards refusing to complete a boot as well.

I'll let you in on a REAL weird one: Since the Parhelia has not worked out on my 7KDD, I went ahead and purchased an ATI AIW 8500: I also had a pair of Radeon 7000 PCI 64 MB DDR cards for another project that got cancelled: these are 66Mhz-compliant cards. The Board would post and boot normally: but the PCI AIWs displayed video corruption until I slowed the 64Bit/66MHz PCI Bus back down to 33MHz by moving some cards. Now, everything is fine.

It could very well be the Chipset Silicon, but I'm still holding out that it is a BIOS issue.

Time will tell.

[Ryu Connor]

These are the known bugs with the 760MP and MPX chipset.

762 Specification sheet:

http://www.amd.com/us-en/assets/content_type/white_papers_and_tech_docs/24089.pdf

766 specification sheet (MP):

http://www.amd.com/us-en/assets/content_type/white_papers_and_tech_docs/23615.pdf

768 specification sheet (MPX):

http://www.amd.com/us-en/assets/content_type/white_papers_and_tech_docs/24472.pdf

Any issues you experiencing that are not specificed in those documents are either motherboard manufacturer specific BIOS bugs, driver issues, design issues, or errata that has not yet been documented.

Multimedia Man's Parhelia problem sounds like a Chaintech BIOS bug.

His RADEON problem on the other hand is probably a design issue that is ATI's fault. The cards might have been keyed for 32/66, but that doesn't necessarily mean they'll work.

[K6-III]

Yeah, sounds like a Chaintech bug to me as well....

[Greebe]

MMM, Matrox and myself (to name a few) are well aware of the problems surrounding AMD's 760mpx chipset... and there are more issues than just what you have posted depending on which chipset revision used.

[DGhost]

I was gonna make a comment on the use of a Chaintech dual processor board... especially since i have worked with some of their other boards...

we had some problems with the MSI MPX board when using an Adaptec card in 64bit/66mhz mode, until we flashed the bios onto it.

i have heard of some AGP problems with the MP/MPX, nothing specific, just that it has some. i would (at this point in time) rather have an AMD chipset than a VIA chipset tho.

i do hope that the AMD Hammer chipsets are as good as they are hoping for... especially as they are trying to get into the enterprise sector... i do hope they continue shipping the hammer chipsets, as i don't like the other choices available...