voidar wrote: I wonder.. Have anyone tried disabling Hardware acceleration on i915 systems?

Voidar,

there is no need to do that. Intel 915 and 925 systems work just as well as 865, 875 systems. AMD PCIe nVidia 4 is a hole different story.

Two DAW software vendors have certified our PCIe systems with remarks as the following;

1). "Hello Guy, I'm happy to report everything has been great with the X2i system. I've been trying all kinds of things with it, and it hasn't given me trouble once."

2). "Guy thank you for sending the X2 for evaluation, you are right it does perform better than the previous 875 system. It rocks!"

3). Hi Guy, this X2i system is a real treat.

Some of the bad publicity PCIe 915/925 is getting is just BS.

Guy Cefalu
SONICA-X, LLC.

SONICA-X wrote: [quote=voidar]I wonder.. Have anyone tried disabling Hardware acceleration on i915 systems?

Voidar,

there is no need to do that. Intel 915 and 925 systems work just as well as 865, 875 systems. AMD PCIe nVidia 4 is a hole different story.

Two DAW software vendors have certified our PCIe systems with remarks as the following;

1). "Hello Guy, I'm happy to report everything has been great with the X2i system. I've been trying all kinds of things with it, and it hasn't given me trouble once."

2). "Guy thank you for sending the X2 for evaluation, you are right it does perform better than the previous 875 system. It rocks!"

3). Hi Guy, this X2i system is a real treat.

Some of the bad publicity PCIe 915/925 is getting is just BS.

Guy Cefalu
SONICA-X, LLC.

Thank you for respodning,

I see that your X2i system uses a very low-level GPU which is adviced by Scott here, so you two do agree on that level.

The big problem supposedly is with PCIe systems with high-level GPU's like x700, x800 etc. I am interested in hearing reports on such setups, and if they perform worse I would like to hear reports on whether disabling HW accel. and/or downclocking the GPU will have an effect on performance.

For what is worth here is my experience;

Based on the information obtained from this and the nuendo forums I built the amd64 3500+ configuration.

Gigabyte K8NS Ultra 939 mobo
amd athlon 939 3500+ 2.2 ghz
1gb ddr400 corsair ram in dual channel
gigabyte nVidia fx5200 agp8x video card
wd 80gb ata-100 system drive
maxtor 300gb sata audio drive
win xp pro svp2 with ms firewire patch
rme fireface800 latest driver and firmware
nuendo 3

I downloaded the thonex test as suggested in the nuendo forum and I can't run it a 6ms latency. Audio clicks and distorts and then the computer stops responding. The cpu meter is always near 90% and when I hit play it goes to 100.

On my dad's computer a PCI express 915P system with a 3.4ghz processor and 512mb ram I can run the test with the latency set to 6ms and I don't get any clicks. the cpu reads 65% and when I hit play goes to 90.

I'm going to return the parts to Newegg and get a 925 pcie system.

Andrew.

Hey Cleo,
sorry to hear that
perhaps you had the Sata drive on the SI chipset rather than the Nvidia. the SI chipset can cause pops clicks randomly.

having tried both myself we get better numbers on the 939.
suggest you move the Sata drive, or there is something else wrong.

on a side note the same system with Giga 3 will get 600 voices.
(2 300G Sata in raid as samples drive)

Scott
ADK

HI Voidar,

I see that your X2i system uses a very low-level GPU which is adviced by Scott here, so you two do agree on that level.

yes on that we agree.

The big problem supposedly is with PCIe systems with high-level GPU's like x700, x800 etc. I am interested in hearing reports on such setups, and if they perform worse I would like to hear reports on whether disabling HW accel. and/or downclocking the GPU will have an effect on performance

.

the faster the video card the worse it becomes for audio.
even an X600 (which is what i used in my first tests)
are dramatic to the point of unuseable.

so again
AMD and PCI-E bad very bad,
Intel 915/925 and PCI-E with the X300 /lower fx good
intel PCI-E and mid to higher video Bad very bad.

Scott
ADK

ADK audio wrote: Hey Cleo,
sorry to hear that
perhaps you had the Sata drive on the SI chipset rather than the Nvidia. the SI chipset can cause pops clicks randomly.

having tried both myself we get better numbers on the 939.
suggest you move the Sata drive, or there is something else wrong.

on a side note the same system with Giga 3 will get 600 voices.
(2 300G Sata in raid as samples drive)

Scott
ADK

ADK,

SATA0_SB is the connector for the maxtor drive which is the nvidia connector.

Also I get way more than a random click. I can't run the test with 6 ms latency. Cpu metter goes to 100 and it becomes unstable.

Run system diagnostics and everything tests OK.

Also the Thonex test doesn't use the hard disk is all VSTI.

I will give it one more day and I am going Intel.

ADK audio wrote: HI Voidar,

I see that your X2i system uses a very low-level GPU which is adviced by Scott here, so you two do agree on that level.

yes on that we agree.

The big problem supposedly is with PCIe systems with high-level GPU's like x700, x800 etc. I am interested in hearing reports on such setups, and if they perform worse I would like to hear reports on whether disabling HW accel. and/or downclocking the GPU will have an effect on performance

.

the faster the video card the worse it becomes for audio.
even an X600 (which is what i used in my first tests)
are dramatic to the point of unuseable.

so again
AMD and PCI-E bad very bad,
Intel 915/925 and PCI-E with the X300 /lower fx good
intel PCI-E and mid to higher video Bad very bad.

Scott
ADK

Thak you for responding.

I have read up on this. But did you by any chance try to experiment with HW accell and/or downclocking the higher-level GPUs? Or does it occupy the same bandwith regardless?

Have you tested disabling un-needed hardware on laptop systems, like mini-pci devices on the Clevo D900T i.e..

considering all the testing we do all the posts i have posted
and that we build for a living yes of course we have tried.
if i tried powerstripe and bios hacking i surely would have tested the obvious.

Scott
ADK

ADK audio wrote: considering all the testing we do all the posts i have posted
and that we build for a living yes of course we have tried.
if i tried powerstripe and bios hacking i surely would have tested the obvious.

Scott
ADK

Ok.

I just wanted to know what had been tested except for just putting system components together. Thank you.

cleo745 wrote: [quote=ADK audio]Hey Cleo,
sorry to hear that
perhaps you had the Sata drive on the SI chipset rather than the Nvidia. the SI chipset can cause pops clicks randomly.

having tried both myself we get better numbers on the 939.
suggest you move the Sata drive, or there is something else wrong.

on a side note the same system with Giga 3 will get 600 voices.
(2 300G Sata in raid as samples drive)

Scott
ADK

ADK,

SATA0_SB is the connector for the maxtor drive which is the nvidia connector.

Also I get way more than a random click. I can't run the test with 6 ms latency. Cpu metter goes to 100 and it becomes unstable.

Run system diagnostics and everything tests OK.

Also the Thonex test doesn't use the hard disk is all VSTI.

I will give it one more day and I am going Intel.

Cleo745,

You are correct! the Thonex test will not create any HDD transfers on the PCI bus. That is not the problem. The problem is that you are running out of CPU power to run this test.

Do me a favor. If you get your Intel system run the test again, come back and post your results.

Get a 3.4Ghz P4/LGA775/1MB which sells for about the same amount as the AMD64 3500+.

Guy Cefalu
SONICA-X

Cleo745,

take a look at this thread from the Nuendo forum;

http://forum.nuendo.com/phpbb2/viewtopic.php?t=121&start=50

These benchmarks have been entered by the same two people who earlier on this thread acused me of hyperbole, etc, etc.

Read the benchmarks very carefully!

Note how the Intel 3.2GHz PCI Express system can run the Thonex test with 256 buffers 6 ms latency without any clicks or pops.

* 256 - Stopped : 62% : Running 70-82 %

Also, note how the AMD64 939 pin 3500+ at 256 buffers 6 ms latency developed clicks and pops.

* 256 very slight pops..

Now from Newegg;

P4 3.2GHz LGA775 Prescott $219.99

ADM64 2.2 939 pin 3500+ $272.00

So, it looks like the new Intel PCI Express systems are not that bad after all!

Go and get your 3.4 PCIe Intel system as I recomended.

I rest my case.

Guy Cefalu
SONICA-X, LLC.

Sonica-X,

If you would take the time to tabulate all qualifying test results (5 Tower, Thonex, orangine...) presented at the Nuendo hardware forum and RME forum you would come to quite a different conclusion, IMHO.

There are at least three different user reports verifying click-free Thonex test results with the AMD 64/939 at 6 ms latency setting (or even at 3 ms). If you would further take the time to adjust users' reported soundcard latency setting values to real latency you would find that - based on the Thonex test and user reports - an AMD 64/939/2.2 GHz o/c to 2.7 GHz - is on par with stress test-latency results with the Intel 925XE/Abit AAX8/P4/640/3.2 GHz o/c to 4 GHz - with a *low-performing PCI-e video card*.

Given the construction of the Thonex test it does not come as a surprise that test results scale positively with increaing cpu frequency - Intel or not.

I'm not interested in any Intel-AMD argumentation per se. I just find that there is too much sales-pitch driven interpretation of DAW stress test results around, and that tests and adjoined interpretation should be conducted in a concise manner as found essential in the scientific community to be meaningful to DAW end-users potentially investing in new technology.

Cheers,
Anders Fahlén

I can say that my older P4 2.8E/875 system (OC'ed to 3.01GHz and with an ATI 9200 8x AGP video card AND with an Internet install of XP-Pro) runs the Thonex test at around 75% @ 6ms with very minor (I mean VERY minor) pops and clicks (playing) with an M-Audio Audiophile 24/96 PCI card. I did not test my "Daw Optimized" install of XP-Pro on this system (yet).

While my newer 775 P4 3.0E / 915G (w/ onboard graphics and the SAME RAM as the 875 system) had severe pops&clicks at 6ms while playing through an RME Multiface (with a DAW optimized XP-Home install), and ran all the way to 90% at times!

I'm not sure exactly how much impact the onboard video has on the 915G/775 system, but the extra crap running on the 875/478 system (for the Internet installation - Firewall, Spybot, etc, etc) should have negated any performance gains from having a 8x AGP video card IMO. Maybe not? I also know the 875/478's light OC (to 215FSB) also helped its RAM bandwidth surpass the 915G/775's stock 200FSB clock (but with slightly relaxed CAS latency for the OC). Not quite apples to apples, but a comaprison none the less...

Anyway, just posting my results with my 2 whacko systems :) ... An ATI x300 PCI-Express 16x card may be in my future for the 915G/775 system. This would appear to put my 2 systems on more of a level playing ground (minus the extra RAM bandwidth on the OC'ed 875/478 system). I'll post results once that purchase happens.

Q: Will the ATI x300 PCI-Express 16x video card help my newer 915/775 system run more efficiently? They only run about $75 - so it is not a big financial risk. Just curious on what to expect with this "new fangled" PCI-Express and my RME Multiface PCI interface... PS - I use all 16 inputs on the Multiface, too!

:cool:

fahlen wrote: Sonica-X,

If you would take the time to tabulate all qualifying test results (5 Tower, Thonex, orangine...) presented at the Nuendo hardware forum and RME forum you would come to quite a different conclusion, IMHO.

There are at least three different user reports verifying click-free Thonex test results with the AMD 64/939 at 6 ms latency setting (or even at 3 ms). If you would further take the time to adjust users' reported soundcard latency setting values to real latency you would find that - based on the Thonex test and user reports - an AMD 64/939/2.2 GHz o/c to 2.7 GHz - is on par with stress test-latency results with the Intel 925XE/Abit AAX8/P4/640/3.2 GHz o/c to 4 GHz - with a *low-performing PCI-e video card*.

Given the construction of the Thonex test it does not come as a surprise that test results scale positively with increaing cpu frequency - Intel or not.

I'm not interested in any Intel-AMD argumentation per se. I just find that there is too much sales-pitch driven interpretation of DAW stress test results around, and that tests and adjoined interpretation should be conducted in a concise manner as found essential in the scientific community to be meaningful to DAW end-users potentially investing in new technology.

Cheers,
Anders Fahlén

Anders,

I don't agree!

I am not interested in test results that come from different users using different sets of hardware in an uncontrolled environment, and I am not interested in test reports that come from overclocked systems.

Also, you are 100% correct,

"interpretation should be conducted in a concise manner as found essential in the scientific community to be meaningful to DAW end-users potentially investing in new technology"

but, the feedback I get from customers is that they believe that PCIe does not work for audio, and that is not true.

The information they are getting comes form forums like this.

See, sometimes people only scan threads and they pick-up only the negative.

This is one example;

RME Multiface

"A comparative reading of audio stress test results from the Intel 915/925 and the NF4 platforms also shows that the performance penalty is less in combination with low-performing PCI-E video cards but still significantly reduced than compared with the NF3/Intel 875-AGP ditto"

I have spoken to customers that believe that PCIe just does not work for audio AMD or Intel just by reading your article.

On the other hand we have customers that are having wonderful results using PCIe systems. Synthax USA and Cakewalk are two of them.

Again, and like the others you can accuse me of creating a sales pitch arround Intel PCIe but that is not the case. We sell both Intel and AMD systems.

Here comes the big question.

If you and others are doing such a good job at interpreting and conducting tests in a concise manner as found essential in the scientific community to be meaningful to DAW end-users potentially investing in new technology,

and

a 3.2GHz PCIe Intel based DAW performs just as well as a 3.2GHz 875P based DAW,

then why are potential customers getting scared of buying PCIe systems?

Should you and the others take the time to conduct tests in a scientific manner or should you take the word of others write an article and post it on the RME site?

And by the way, just by the nature of the Thonex test and the fact that all this data comes from different sources with different test beds in an uncontrolled test environment makes it a non scientific test.

Perhaps a DAW vendor without any ties to a CPU manufacturer should pick this up in a QA lab?

My best.

Guy Cefalu.

SONICA-X,

If you would have bothered to quote the next sentence from the referred RME article you would maybe have understood why none - bar none - has neither questioned nor been left in the dims regarding possible working DAW combination of Intel 9xx - properly configured - on the RME forum. This is because we have had a long constuctive thread related to PCI-e, NF4 and Intel 9xx , largely devoid from sales pitch. If you would have followed these threads you could have saved yourself from bringing this one up. Fellow forum members with Intel 9xx platforms as well as NF4 have constructively added to the knowledge record as of know.

On the other hand, users have also responded with reports on other poential quirks with the Intel 9xx under specific settings as well as with the NF3/NF4 - not PCI-e related. Intel or not - my view is that such qualifying reports are valid as peer-to-peer information as to (1) allow for troubleshooting diagnosis and possible resolutions and (2) to make DAW users alert on *potential* hardware and configuration issues. From responses so far, I reckon that some 10+ Intel 9xx users have been saved from further troubles by bridging information from the referred article to their own troubleshooting situation. The RME article has also lead to an interest to have a deeper look lnto possible NF4 PCI-e/PCI interconnectivity issues by independent hardware analysist. We'll see what comes - in the longer run I'm quite confident that PCI-e will be mainstreamed into the DAW community with robust performance irrespective chipset/cpu preferences.

I have no interest to engage myself any further into the rather bizarre and personal discussions covering most of this thread. And no - personally I have no hidden agenda (Intelmania or other) than to try to add peer-to-peer services to DAW users. The verdict is that both platforms work well, properly configured including the Intel 9xx chipset, under a range of DAW workflows but the used techniques to partially single out DAW stress test results to proof "A" or "B" is rather meaniingless as long as the result is not compiled and interpreted in a way where chance is separated from true differences.

In the end, even a Thonex test concisely managed is just one way of trying to measure robust performance on a specific DAW platform and should not be looked upon as *the only criteria* from an end-user's perspective.

Cheers,
Anders Fahlén

Would I be likely to experience the same problems with an nForce4 based board and a USB or Fiurewire audio card? Thanks

HI ,

Yes the PCI-E will affect any type of I/O.

It also depends on track count, effects count, sample rate, VSTi etc
Someone doing a few tracks (8-16) and little of anything else I am sure would be able to use
The PCI-E Nforce.

The tests we use are very heavy hitters which separate the men from the boys (computer wise)
Basically we came up with some tests that would drop a 2 yrs old computer to its knees.

I have a brand new AMD laptop, gig of ram AMD 64 3400 (754 pins) 60G 7200 HDD and 1 test I have it wont run it past 24ms.

Scott
ADK

Scott... I have the Sonar3Test
I have used AMD CPUs before and currently have P4 650 i925x mobo.
Where do I post the results?
Thanx

Strings

Hey Strings,

thats a good question. there is a thread on Nuendo for that test and there is a thread on Sonar for the Old (Scott Reams) sonar test. but there is not a thread for the new ADK sonar test.

i assume thats the one you used.. as the old one doesnt give a good indication of a systems true ability.

here would be good i guess.

hmm i could start one on the Sonar forum as well.

Scott
ADK

Thanx Scott

I got the file a while ago from RoadRec from an article on Sonar 3.1 re multiprocessing.
I'm using Sonar PE 3.1.1
Do you want those results?

Strings

HI Strings,

a better test is here
http://recording.org/resources/recording-computers.300/

the ADK Sonar test

thanks
scott