VOGONS


Reply 20 of 38, by feipoa

User metadata
Rank l33t++
Rank
l33t++
PCBONEZ wrote:
. PCI-X like conventional PCI is a shared bus. If the specific PCI-X slot you used happens to be shared by something else on the […]
Show full quote
feipoa wrote:

With the Parhelia in the PCI-X slot, the score was 3854. With the Parhelia in the conventional PCI slot, the score was 4580. Any ideas why the score would be higher with the card in the conventional PCI slot?

.
PCI-X like conventional PCI is a shared bus.
If the specific PCI-X slot you used happens to be shared by something else on the motherboard (drive controllers and LAN are common) then the full bandwidth is not available to the video card.

It is not uncommon for there to be two or even three independent PCI-X buses on a motherboard.
If you want to use PCI-X video it would be a good idea to find out which PCI-X bus has the least on it.
Manuals for boards with PCI-X usually show what is on what bus in the manual.
.

That is a good point. I no longer recall if I looked into that or not. I did these tests some time ago. Let's assume that the gigabit ethernet adapter and the Matrox graphics card are both on the PCI-X bus. If the ethernet adapter is enabled, but not in use (ethernet cable not plugged in), to what extent does this effect the throughput of the Matrox graphics card? If the terhnet adapter is disabled in the BIOS, to what effect does this have on the Matrox graphics card?

Plan your life wisely, you'll be dead before you know it.

Reply 21 of 38, by PCBONEZ

User metadata
Rank Oldbie
Rank
Oldbie

I attached a snap-shot to to my previous post. (Noted as the page has 'turned' here.)

If the LAN chip is disabled it won't be using any bandwidth. A little power maybe, but no bandwidth.
.

GRUMPY OLD FART - On Hiatus, sort'a
Mann-Made Global Warming. - We should be more concerned about the Intellectual Climate.
You can teach a man to fish and feed him for life, but if he can't handle sushi you must also teach him to cook.

Reply 22 of 38, by feipoa

User metadata
Rank l33t++
Rank
l33t++

I recall when using a USB card in my PCI-based 486 system, just having the USB card enabled, although not in use, lowered some benchmark scores by 5-10%.

In my above tests, I am fairly confident that I left the gigabit ethernet enabled and was using the onboard SCSI, which may or may not have been on the same bus. It would be interesting to rerun the benchmarks to determine how much bandwidth is consumed for other idle components on the PCI-X bus.

Plan your life wisely, you'll be dead before you know it.

Reply 23 of 38, by PCBONEZ

User metadata
Rank Oldbie
Rank
Oldbie

Another possibility is that some PCI slots that look conventional are actually 66MHz capable.
Those never got super popular but there are some around.
.

Last edited by PCBONEZ on 2016-01-08, 09:02. Edited 1 time in total.

GRUMPY OLD FART - On Hiatus, sort'a
Mann-Made Global Warming. - We should be more concerned about the Intellectual Climate.
You can teach a man to fish and feed him for life, but if he can't handle sushi you must also teach him to cook.

Reply 24 of 38, by PCBONEZ

User metadata
Rank Oldbie
Rank
Oldbie
feipoa wrote:

I recall when using a USB card in my PCI-based 486 system, just having the USB card enabled, although not in use, lowered some benchmark scores by 5-10%.

Makes sense because there is a little chatter between the USB chip and the system even when no USB device is attached.
I think most of it is the system checking to see if you just plugged something into USB.
.

GRUMPY OLD FART - On Hiatus, sort'a
Mann-Made Global Warming. - We should be more concerned about the Intellectual Climate.
You can teach a man to fish and feed him for life, but if he can't handle sushi you must also teach him to cook.

Reply 25 of 38, by feipoa

User metadata
Rank l33t++
Rank
l33t++

How can you pass 64-bits of bus data on a non-PCI-X slot? I thought there were only 32-bit/66 MHz and 32-bit/33 MHz conventional PCI slots?

Plan your life wisely, you'll be dead before you know it.

Reply 26 of 38, by PCBONEZ

User metadata
Rank Oldbie
Rank
Oldbie

Yes, I fixed that.

Last edited by PCBONEZ on 2016-01-08, 15:05. Edited 1 time in total.

GRUMPY OLD FART - On Hiatus, sort'a
Mann-Made Global Warming. - We should be more concerned about the Intellectual Climate.
You can teach a man to fish and feed him for life, but if he can't handle sushi you must also teach him to cook.

Reply 28 of 38, by swaaye

User metadata
Rank l33t++
Rank
l33t++
dr.zeissler wrote:

I have one to sell. Shipping EU only (no German Zoll)
It is silent with active cooling.

Please note that we don't allow people to sell items on VOGONS.
VOGONS is no marketplace!

Reply 30 of 38, by gdjacobs

User metadata
Rank l33t++
Rank
l33t++

Although this likely didn't come into play in this case, it's important to note that PCI devices which don't support higher PCI-X clocking (66, 100, 133mhz) will limit the speed of the overall bus. This is another reason to pay attention to bus mapping on server motherboards.

All hail the Great Capacitor Brand Finder

Reply 31 of 38, by idspispopd

User metadata
Rank Oldbie
Rank
Oldbie

Would the Windows Device manager tell you which PCI device is on which bus? XP already has "Devices by Connection" in the device manager.

Reply 32 of 38, by PCBONEZ

User metadata
Rank Oldbie
Rank
Oldbie
idspispopd wrote:

Would the Windows Device manager tell you which PCI device is on which bus? XP already has "Devices by Connection" in the device manager.

It should. It does for PCI and PCIe.

If I can find a hoist big enough to alleviate my rectal cranial inversion problem I'll check tomorrow.
Might not work out as all the PCI-X cards I can think of that I have here are 3Ware RAID cards and they don't like to play with the Adaptec RAID in the PCIe slot.
.

GRUMPY OLD FART - On Hiatus, sort'a
Mann-Made Global Warming. - We should be more concerned about the Intellectual Climate.
You can teach a man to fish and feed him for life, but if he can't handle sushi you must also teach him to cook.

Reply 33 of 38, by mzry

User metadata
Rank Member
Rank
Member

Sorry to resurrect an old thread, but I managed to get hold of a Parhelia agp 8x card, which had the rumored new chip revision and better performance. I can confirm that this is indeed the case, the chip/package size is almost half compared to the original parhelia, here is a photo:
https://drive.google.com/open?id=1geBpSN3kK93 … X0Nkn4SU6ro3Pq1
original version for size comparison:
https://techreport.com/r.x/parhelia/parhelia-chip.jpg

System: Asus tusl / Tualatin 1.4 / 512mb infineon cas2 ram / win2k / driver: 1.13.0.158
3dmark 2001 default benchmark results: 7401
Quake 3 @ 1024x768 32bit no fsaa: 101fps

This is 1000 points higher in 3dmark and 20fps higher in Q3 compared to my retail version 1 parhelia. I haven't attempted to overclock it yet, but the original parhelia didn't like to OC at all. Apparently this one does.

ps another interesting tidbit: using this latest driver and this card, dxdiag reports it as being dx9. But the parhelia was originally shamed for not supporting 9, but 8 only.

Last edited by mzry on 2018-02-10, 12:04. Edited 1 time in total.

Reply 35 of 38, by Sphere478

User metadata
Rank l33t++
Rank
l33t++
mzry wrote on 2018-02-09, 10:14:
Sorry to resurrect an old thread, but I managed to get hold of a Parhelia agp 8x card, which had the rumored new chip revision a […]
Show full quote

Sorry to resurrect an old thread, but I managed to get hold of a Parhelia agp 8x card, which had the rumored new chip revision and better performance. I can confirm that this is indeed the case, the chip/package size is almost half compared to the original parhelia, here is a photo:
https://drive.google.com/open?id=1geBpSN3kK93 … X0Nkn4SU6ro3Pq1
original version for size comparison:
https://techreport.com/r.x/parhelia/parhelia-chip.jpg

System: Asus tusl / Tualatin 1.4 / 512mb infineon cas2 ram / win2k / driver: 1.13.0.158
3dmark 2001 default benchmark results: 7401
Quake 3 @ 1024x768 32bit no fsaa: 101fps

This is 1000 points higher in 3dmark and 20fps higher in Q3 compared to my retail version 1 parhelia. I haven't attempted to overclock it yet, but the original parhelia didn't like to OC at all. Apparently this one does.

ps another interesting tidbit: using this latest driver and this card, dxdiag reports it as being dx9. But the parhelia was originally shamed for not supporting 9, but 8 only.

Sorry for the necro, but any chance of fixing those dead image links? Maybe a host here?

Just got one of these, was thinkin about shoehorning it into a dual pentium 233 build to see what happened.

Tryin to beat 9250 score on the same system. Sounds like by Feiopa’s account the 9250 is faster though. In which case I won’t bother, it’s a cool collection piece if nothing else. But if I have a better version, may give it a go.

Btw, no win 9x drivers avail? Right?

Sphere's PCB projects.
-
Sphere’s socket 5/7 cpu collection.
-
SUCCESSFUL K6-2+ to K6-3+ Full Cache Enable Mod
-
Tyan S1564S to S1564D single to dual processor conversion (also s1563 and s1562)

Reply 36 of 38, by Trashbytes

User metadata
Rank Oldbie
Rank
Oldbie
Sphere478 wrote on 2024-03-12, 01:43:
Sorry for the necro, but any chance of fixing those dead image links? Maybe a host here? […]
Show full quote
mzry wrote on 2018-02-09, 10:14:
Sorry to resurrect an old thread, but I managed to get hold of a Parhelia agp 8x card, which had the rumored new chip revision a […]
Show full quote

Sorry to resurrect an old thread, but I managed to get hold of a Parhelia agp 8x card, which had the rumored new chip revision and better performance. I can confirm that this is indeed the case, the chip/package size is almost half compared to the original parhelia, here is a photo:
https://drive.google.com/open?id=1geBpSN3kK93 … X0Nkn4SU6ro3Pq1
original version for size comparison:
https://techreport.com/r.x/parhelia/parhelia-chip.jpg

System: Asus tusl / Tualatin 1.4 / 512mb infineon cas2 ram / win2k / driver: 1.13.0.158
3dmark 2001 default benchmark results: 7401
Quake 3 @ 1024x768 32bit no fsaa: 101fps

This is 1000 points higher in 3dmark and 20fps higher in Q3 compared to my retail version 1 parhelia. I haven't attempted to overclock it yet, but the original parhelia didn't like to OC at all. Apparently this one does.

ps another interesting tidbit: using this latest driver and this card, dxdiag reports it as being dx9. But the parhelia was originally shamed for not supporting 9, but 8 only.

Sorry for the necro, but any chance of fixing those dead image links? Maybe a host here?

Just got one of these, was thinkin about shoehorning it into a dual pentium 233 build to see what happened.

Tryin to beat 9250 score on the same system. Sounds like by Feiopa’s account the 9250 is faster though. In which case I won’t bother, it’s a cool collection piece if nothing else. But if I have a better version, may give it a go.

Btw, no win 9x drivers avail? Right?

Correct no 9x drivers, the Parhelia is really not suited for gaming, its perfect for industrial purposes or medical purposes where its used in imaging systems requiring exceptional image quality but gaming .. yeah you might want to go with the 9250.

Reply 37 of 38, by Grem Five

User metadata
Rank Member
Rank
Member

I cant compare my former test to Feiopa's test as it sounds like it was not done using default settings but Re: A useless short analysis on PCIX to PCIe bridge performance on Tualatin

I believe what mzry was referring to was the agp versions like here but most of that was down to higher clocks on later versions.

I know they made 2 versions of the Parhelia 256 PCI-X as mine is different than most of the ones I see pictured on the net, mine seems to be an earlier version but if I remember right it has the same clocks or really close to the later agp 8x versions.

Its late here maybe I can post more details tomorrow.

Reply 38 of 38, by Grem Five

User metadata
Rank Member
Rank
Member
Sphere478 wrote on 2024-03-12, 01:43:

Tryin to beat 9250 score on the same system. Sounds like by Feiopa’s account the 9250 is faster though. In which case I won’t bother, it’s a cool collection piece if nothing else. But if I have a better version, may give it a go.

In this other thread Re: High end PCI video card for Tualatin? it seems Feiopa has the Matrox beating out the 9250 in two out of the three tests but not by much.

Judging by the specs listed there my Parhelia is clocked slightly faster on the core clock but the same on memory. I dont know if this is the difference but most of the Parhelia 256 PCI cards I have seen are board revision 'A' while mine is a revision 'B'. Most of the times I see it listed that the Parhelia 256 PCI had a core clock of 220 with memory clock of 275 and mine is core clock of 250 with the same memory clock of 275 so just a small difference of core clock.

I'm kinda surprised they run the memory so low as almost all the ones I have seen have 2.8 ns memory which should be fine running near 350.

Although I dont know if the revision makes any difference as mine is a 7128-0101 Rev B and the other cards I have seen with a Rev A have been 7128-0102 Rev A and 7128-0103 Rev A. I'm unsure of the other markings I have seen, mine is a MGI PH-P256 but I have also seen MGI PH-P256F and MGI PH-P256PDPIF.

Pic of my card is here Re: What is PCI-X?

Unless you are going to put it in a 64bit 66 MHz pci slot probably better off with the 9250. I would do my own testing of my 9250 vs this card but to lazy for now.