Just thought I would let you all know, I recently got a tyan tiger mp I'm running 1 1.4ghz athlon mp, win98, 512mb ram, a pulsar II and a luna and I get 11 masterverbs which maxes out my dsp before any pci overflows (with 24track 24bit playing in sonar). If ne1 (sub) knows of a better pci test let me know. I played with pci latiency timer in the bios and changed the setting to the one on the bottum of the list (hexadecimal) for the card slot where the pulsar II is. I also bought a promise ultra133tx2 which is 64 bit PCI so I can see if it makes a difference (when it gets here). Another Q: is XP limited to 12 inputs 8 outputs or is that 2k? and one more: anyone here try those
3ware raid 10 systems? thanks
Ryan
Sub 11 masterverbs with tyan tiger mp!!
Excellent results!! You've got me excited again, but I swear we tried that exact same thing, so you used a LOWER pci Latency on your card? Which slot number are your cards in -- the 32bit slots, or the 64bit slots? We tried every single one, and a ton of PCI latency numbers -- so you're using:
1. Windows XP with ACPI Multi Proc(?)
2. Creamware cards with PCI Latency at 20Hex (lowest)?
3. Any patches applied...?
I'm not sure how to tell the differences between the various board revisions. We had one of the first boards to hit the states, so it could be we just got a bad board.
Exciting news to say the least!! Now, were you showing DSP load, and did you have your Creamware cards showing clock? Because you can load this many on even a VIA chispet, if your DSPs aren't clocked...
Also try running audio through them (ie: from the ADAT inputs or something).
1. Windows XP with ACPI Multi Proc(?)
2. Creamware cards with PCI Latency at 20Hex (lowest)?
3. Any patches applied...?
I'm not sure how to tell the differences between the various board revisions. We had one of the first boards to hit the states, so it could be we just got a bad board.
Exciting news to say the least!! Now, were you showing DSP load, and did you have your Creamware cards showing clock? Because you can load this many on even a VIA chispet, if your DSPs aren't clocked...
Also try running audio through them (ie: from the ADAT inputs or something).
The card is clocked (slaved to adat) and audio runs through (sounds really muddy but cool as hell) this is while playing back 24tracks 24bit 44.1. my guess is that without the tracks playing and more DSP's I could get even more.
Ok here's whats strange but really cool. . .I accedentally adjusted latency timer of slot 3 to 00e0h at the bottum (this should be the highest) but I looked again and my card is in slot 2 (64bit slot). after adjusting them all to default 0040h it tends to crash and get buggy but still no pci overflows, but really high cpu usage. . .with slot two set to 00e0h its not much better but, slot 2 (pulsar II) set to 0020h (lowest Pci latency right?) and slot 3 set to 00e0h (highest?) and its about the same. so heres the lowdown I can run, it seems, as much over the pci bus (with any different latency settings) as I want and the total system latiency is increased in order to wait for >133mb/s (thus decreasing processor effeciency) to move accross the bus to the north bridge. apparently latecny timer has no effect??? (now pulsar is set to 0020h) but CPU usage and drive usage meters are 20-50% higher with a bogged down pci bus. my guess is that the 64bit adapter will clear this up even more (if so I'm going to switch to a raid 10). but lukily I have no projects that are strictly 11 materverbs. I can even run 4 pauls verbs (not even one would run on an older 440bx)
The board is running bios 1.03 (the only one available) I'm using pulsar v3.00, agp card and a cheap nic. no special patches or otherwise. . .
I'm still running 1 proc with win 98 (I like to take these upgrades 1 at a time so if somthing goes wrong I know where to go back to). I looked all over the board for a rev # but I don't see anything
. but hey guess I just got lucky. . .anyway I'll keep posting my results as I move to xp and get the othor proc and more ram and the 64bit ata card.
P.S. if you ever have the chance to try this many masterverbs at once the sound is so cool. . not even a chance you could tell what sound went in.
also sub do you know about the io setup in xp? is it only 12/8 or more?
Ok here's whats strange but really cool. . .I accedentally adjusted latency timer of slot 3 to 00e0h at the bottum (this should be the highest) but I looked again and my card is in slot 2 (64bit slot). after adjusting them all to default 0040h it tends to crash and get buggy but still no pci overflows, but really high cpu usage. . .with slot two set to 00e0h its not much better but, slot 2 (pulsar II) set to 0020h (lowest Pci latency right?) and slot 3 set to 00e0h (highest?) and its about the same. so heres the lowdown I can run, it seems, as much over the pci bus (with any different latency settings) as I want and the total system latiency is increased in order to wait for >133mb/s (thus decreasing processor effeciency) to move accross the bus to the north bridge. apparently latecny timer has no effect??? (now pulsar is set to 0020h) but CPU usage and drive usage meters are 20-50% higher with a bogged down pci bus. my guess is that the 64bit adapter will clear this up even more (if so I'm going to switch to a raid 10). but lukily I have no projects that are strictly 11 materverbs. I can even run 4 pauls verbs (not even one would run on an older 440bx)
The board is running bios 1.03 (the only one available) I'm using pulsar v3.00, agp card and a cheap nic. no special patches or otherwise. . .
I'm still running 1 proc with win 98 (I like to take these upgrades 1 at a time so if somthing goes wrong I know where to go back to). I looked all over the board for a rev # but I don't see anything
. but hey guess I just got lucky. . .anyway I'll keep posting my results as I move to xp and get the othor proc and more ram and the 64bit ata card.P.S. if you ever have the chance to try this many masterverbs at once the sound is so cool. . not even a chance you could tell what sound went in.
also sub do you know about the io setup in xp? is it only 12/8 or more?
I think it's 32! But maybe even more... obviously there will need to be some tweaks done to Creamware's drivers to take full advantage of Windows XP... it only just came out but already I know this is the OS I really want to use fulltime.
Our replacement board should arrive in about 10 days, so once we're back up and running, I'll try to duplicate your results and share anything else we find out... BTW, we also have the 1.6MP chips just like you!
Our replacement board should arrive in about 10 days, so once we're back up and running, I'll try to duplicate your results and share anything else we find out... BTW, we also have the 1.6MP chips just like you!
couple of updates,
I get stutters at low ulli and sonar latencies, but one big thing for anyone with a agp card with video capture: the agp bus is only independant of PCI for computer to video card, any data coming back is normal PCI. I.E. if you have an ATI all in wonder or similer compare PCI performance with the capture drivers disabled. for me its less stutters at lower latencies (the new tyan board will NEVER give me a PCI overflow but it can be unusable at low ulli/sonar latency settings). The promise ultra 133 tx2 64bit ata card will be here in a week or so, I'll let everbody know if that free's up the 32 bit bus or not. Anyway right now I'm about a 100% (95%CPU before is about 45%now) increase in performance from my old cusl2/1ghz/384, latency is lower and the GUI is fast enough to bear.
I get stutters at low ulli and sonar latencies, but one big thing for anyone with a agp card with video capture: the agp bus is only independant of PCI for computer to video card, any data coming back is normal PCI. I.E. if you have an ATI all in wonder or similer compare PCI performance with the capture drivers disabled. for me its less stutters at lower latencies (the new tyan board will NEVER give me a PCI overflow but it can be unusable at low ulli/sonar latency settings). The promise ultra 133 tx2 64bit ata card will be here in a week or so, I'll let everbody know if that free's up the 32 bit bus or not. Anyway right now I'm about a 100% (95%CPU before is about 45%now) increase in performance from my old cusl2/1ghz/384, latency is lower and the GUI is fast enough to bear.
I did the test on a 100Mhz bus mac G4 400Mhz. I could run 8 master verbs (not the classic one) with a ULLI setting of 11ms. I was able to load up 10, but then It started using card 2 aswell. How did u manage to run 11 master verbs on one card?
I run the test by dragging the verbs to the project window and then routing ASIO source to all of them and then mix them together using a micromixer.
I'll do the test with the 867Mhz mac I have at home. I'll get back.
I run the test by dragging the verbs to the project window and then routing ASIO source to all of them and then mix them together using a micromixer.
I'll do the test with the 867Mhz mac I have at home. I'll get back.
-
marcuspocus
- Posts: 2310
- Joined: Sun Mar 25, 2001 4:00 pm
- Location: Canada/France
Updates:
OK so I installed XP (dual boot setup) and have some interesting news. I get copy protection violation errors like ne1 else using a SMP mobo with XP. Also after installing the 3.01 update I start to get the old PCI master overflow error. So out of cuiousity I started to mess with the
[DSP]
maxpcchannel:
option and this is where it gets interesting. With lower numbers I can load devices and not get either error(?) but obviously I also cant get as much stuff loaded as I could with a higher number(without getting the "other"[in a windows window] pci overflow message followed by the normal one). with a high number I get both errors (which ever is first) but not really with any consistancy i.e. somtimes one masterverb = overflow and somtimes it's six. certainly 6 masterverbs will not usually load without a copy protection violation (loading 6 masterverbs is like a 1 in 20 possibility). My conclusion is that the XP drivers are far from being "there" with only 8 wave/wdm i/o's and 16bit only (as well as the other problems) this is not a realistic choice for me now. My advice to ne1 considering this system is mabye get the mobo and stick with 98 (and one proc) until creamware releases the next update (or two). Seeing as how things as simple as midi interfaces and stereo sound cards don't have good XP implementation yet I don't blame creamware, the fact that it runs on XP/2k at all is a good sign that they will soon make it actually work(for me anyways). Also I mentioned before that I was going to get a 64bit pci ide card but the promise ultra 133 tx2 is only 32bit and 66mhz(capable) so I hope I havn't missinformed ne1. One question that seems to be going around is what south bridge (if any) i.e. 760mp or 760mpx, allows the other PCI bus it's own connection (independant of the 32bit 33mhz one that the pulsar uses)? I can't seem to find anything on this at all. Also I could not find where to pick between ACPI mode and standard PC(?) mode so I assume that its installed under the ACPI mode which is not recomended (but required for SMP). Well, thats all I know for now, thanks,
Ryan
OK so I installed XP (dual boot setup) and have some interesting news. I get copy protection violation errors like ne1 else using a SMP mobo with XP. Also after installing the 3.01 update I start to get the old PCI master overflow error. So out of cuiousity I started to mess with the
[DSP]
maxpcchannel:
option and this is where it gets interesting. With lower numbers I can load devices and not get either error(?) but obviously I also cant get as much stuff loaded as I could with a higher number(without getting the "other"[in a windows window] pci overflow message followed by the normal one). with a high number I get both errors (which ever is first) but not really with any consistancy i.e. somtimes one masterverb = overflow and somtimes it's six. certainly 6 masterverbs will not usually load without a copy protection violation (loading 6 masterverbs is like a 1 in 20 possibility). My conclusion is that the XP drivers are far from being "there" with only 8 wave/wdm i/o's and 16bit only (as well as the other problems) this is not a realistic choice for me now. My advice to ne1 considering this system is mabye get the mobo and stick with 98 (and one proc) until creamware releases the next update (or two). Seeing as how things as simple as midi interfaces and stereo sound cards don't have good XP implementation yet I don't blame creamware, the fact that it runs on XP/2k at all is a good sign that they will soon make it actually work(for me anyways). Also I mentioned before that I was going to get a 64bit pci ide card but the promise ultra 133 tx2 is only 32bit and 66mhz(capable) so I hope I havn't missinformed ne1. One question that seems to be going around is what south bridge (if any) i.e. 760mp or 760mpx, allows the other PCI bus it's own connection (independant of the 32bit 33mhz one that the pulsar uses)? I can't seem to find anything on this at all. Also I could not find where to pick between ACPI mode and standard PC(?) mode so I assume that its installed under the ACPI mode which is not recomended (but required for SMP). Well, thats all I know for now, thanks,
Ryan
Heh yea I remeber when I was into macs and I always hated that, not to be the Mac vs Pc asshole because their both cool but Microsoft does one thing that I like its their huge emphasis on backward compatibility. Then making the upgrade an extension of a pre-existing line of OS's is brilliant. but after a year or so when OSx gets full support you'll be the one smileing cause it runs on a unix Kernal. . . 6 to 1 half dozen I guess. . .
Hehe, I just tried how much my old crate can handle...
Saturn routed thru 8 masterverbs, I was even able to load no 9 and 10, but the 10th caused the PCI overflow and I had to remove the last two to get rid of it again. So it works stable with 8 masterverbs. But my machine is still that old P3/500 on an ASUS P3B-F (i BX440) with Adaptec 29160 SCSI and Win98. Not bad for this quite small computer compared to the actual machines. ULLI settings were fastest possible at 44.1kHz and PCI latency timer set to 128.
<font size=-1>[ This Message was edited by: JoeKa on 2002-01-20 00:32 ]</font>
Saturn routed thru 8 masterverbs, I was even able to load no 9 and 10, but the 10th caused the PCI overflow and I had to remove the last two to get rid of it again. So it works stable with 8 masterverbs. But my machine is still that old P3/500 on an ASUS P3B-F (i BX440) with Adaptec 29160 SCSI and Win98. Not bad for this quite small computer compared to the actual machines. ULLI settings were fastest possible at 44.1kHz and PCI latency timer set to 128.
<font size=-1>[ This Message was edited by: JoeKa on 2002-01-20 00:32 ]</font>
