Sounds Cards and Via Chipset [Archive]

DCO

10-30-2001, 03:53 AM

Just the name...."VIA" makes my skin crawl. Via chips are garbage. And if you think I'm being nasty, you should hear all the horror stories I have regarding audio, modem, and USB problems that I eventually discovered were due to VIA chipsets.
If you have a PC and use it only for word processing or just to browse the net, fine, save some bucks and get a PC with a Via based mobo. But any serious computing needs should be based on Intel, ESPECIALLY audio.
You could shop around for a sound card that going to work with a badly designed mobo, but you are engineering from a weak point in the chain. You might just consider getting a new mobo with good old Intel chips (CPU is the subject of a whole other post, but here too I swear by Intel). It wouldn't cost any more then getting a new sound card. It might mean more work, but a good tech can switch a mobo in 20 minutes.
I'm not always positive about every opinion I post here, but this one I am.
DCO

papario

10-30-2001, 06:46 PM

Hi DCO:

If the people or ProRec.com and other Net Sites, listen your suggestions about that an Audio PC is INTEL, they will get sick. To my knowledge and why I built my great Athlon 1400 was based on expertise from Mags around the world and specially, Magazines Experts working for the audio Industry. So I am sorry, but you are really wrong. I am not saying that a Pentium is wrong, I am saying is that you are wrong with your comments about the Via chipset and about that an Audio PC has to be Pentium.

The Via chipset KT133 A and better; has no more incompatibility problems. You are talking about the VIA KT 133, but that is a very, very old story. Here I am, With a great system very powerful that never fails, even under stress tests.....

Read more here from the experts:
http://prorec.com/prorec/articles.nsf/articles/D6A94D6B2F7DA99186256A7F00747CC9

Plus, let me add that AthlonXP(the new palomino) is greatly optimized for windowsXP and I have read already that these 2 will be the way to go in the near future for Music Productions.

Greetings
papario

xanto

10-30-2001, 07:01 PM

Here is another happy Athlon user. My PC is only for Audio. No office, no Net, no bul****. Only Audio. It´s the way to go. I agree with papario due I also read that article and others. My Via chipset KT133 A works so great that my system never crashes or whatsover.

xanto

Leadfoot

11-07-2001, 09:25 PM

Well I have to chime in here for VIA.
I've been using a via chipset in one form or another and I think they are the best.
In fact the only time I ever had a problem with a chipset it was an Intel 440bx!
Via is constantly updating their drivers which tells me that they care about always bettering their chipsets. And no that doesn't mean that they are flawed to begin with , it means that they don't forget about stuff they've already sold.
My 2 cents,
Lead.

buddy

11-10-2001, 06:56 PM

Please, i want to know if there is any problem with my mobo Asus k7v (Via KX133)and sounds card like Layla24,Delta 10-10,Oasys,..

knuttern

11-18-2001, 04:40 AM

Originally posted by papario:
Hi DCO:

If the people or ProRec.com and other Net Sites, listen your suggestions about that an Audio PC is INTEL, they will get sick. To my knowledge and why I built my great Athlon 1400 was based on expertise from Mags around the world and specially, Magazines Experts working for the audio Industry. So I am sorry, but you are really wrong. I am not saying that a Pentium is wrong, I am saying is that you are wrong with your comments about the Via chipset and about that an Audio PC has to be Pentium.

The Via chipset KT133 A and better; has no more incompatibility problems. You are talking about the VIA KT 133, but that is a very, very old story. Here I am, With a great system very powerful that never fails, even under stress tests.....

Read more here from the experts:
http://prorec.com/prorec/articles.nsf/articles/D6A94D6B2F7DA99186256A7F00747CC9

Plus, let me add that AthlonXP(the new palomino) is greatly optimized for windowsXP and I have read already that these 2 will be the way to go in the near future for Music Productions.

Greetings
papario

I`m sorry Pappario, but it is you who are wrong.Via have had, and has a lot of problems. kt133a is among the worst. It use the revised 686b northbridge with pci errors that cause everything from dataloss while playing back mp3s to low pci bandwidth causing stuttering and bad audio playback.

The 686b is used by every via chip from kt133a to the newest kt266a. In addition to 99% of all amd 761 based mobos.

It is also worth noting that not all users encounter these problems. It is dependent on the brand of mobo you use, other hardware, os/software and how hard you are pressing its resources.

Check out http://forums.viaarena.com/ If you`d like to see what I mean.

My sad conclusion is that if you want to get prosumer results from any AMD based system these days you either have to:

1. Be very lucky
2. Switch between 10 different mobos and other hardware
3. Be prepared to spend weeks tweaking bios and driver revisions every time you make a change of your system.

Tthis is a real shame because AMD processors are now the best performers as well as best value. What most people want it to plug in the soundcard, install the drivers from cd, and go make music. Via is not able to provide that solution reliably (even though _some_ achieve this)

Best regards
K.

BYTEME

11-18-2001, 08:35 AM

well, let's at least get the facts correct...

the 686b is a southbridge chip, not a northbridge. the data corruption issue is primarily related to cases involving a sb live! card, because in fact the live! card does not completely adhere to the pci bus-master specifications as it should, so this is kind of a finger pointing thing.

the newer bios updates and 4-in-1 drivers fix this issue.

the new kt266 / kt266a chipsets do NOT use the 686b chip, they use the new vt8233 chip with "v-link" 266mb/s transfer rate.

xanto

11-18-2001, 02:59 PM

Hi knuttern.

I totally disagree with you. The post talks about what is recomended by the music experts and not based on what people comments on a forum about their problems. So, do you buy your products based on forums or based on expertise??

By the way, you made a serious error about what is a southbrigde and a northbridge.

When I bought my Athlon was also using the right expertise from the music experts which have of course the right info, they have the labs to suggest what is better or what is not. When I buy something, of course I check the forums to have a different perspective, but I don´t buy NEVER based on what people I don´t know say about this or that in a public forum.

I check the experts, the music mags, the music professionals to tell me, what is better and what is not. If you don´t agree, well, I can´t do anything.

If you read the link to the prorec.com and you don´t believe it, then it is your own problem of course. Since I bought my system, I have never changed to different mobos "as you say" and have never had any problem of any kind using the via chipset kt133a.

You say is the worst, but I prefer to trust the guys with the right tools in a lab to make a decission for me. In fact, I am very happy with my PC and God bless me for buying an Athlon and not a Pentium. I had either any problem using my SB Live and other cards plugged in like TVs, Network,etc.

PS; And by the way, it´s funny to see a new member going directly to discuss on a topic. I remember when I was a new member, I saw so many posts as yours, but prefered to take the time to see what all this about.

xanto

Gremsongs

11-19-2001, 07:47 AM

I'll have to chime in here also. I have read a very small but vocal minority speaking against Via chipsets as if they were personally affected by them. And most that I have read about don't/won't/never use Via chipsets. So most talk with no authority. And the response to that comment I just made is always "I'll never use a Via Anything" or "I base my decision on ..." and then quote some authority that is at the very least questionable.

So I don't pay much attention to these guys. I use Via chipsets for over a 2 years now. And any problem I have had, has been corrected by a driver update from the hardware that is in the computer. Or it was a Windows issue. But not with the chipset. These guys always say it is the Via chipset's incompatiblity issue.

But the bottom line is that I used a BX mother board for over two years and had some of the same problems as I have now. And I know that the problems that I encountered are software related and not the Via chipset.

I am not anti-Intel. I'm just going to use whatever is the best for the money I spend. Biggest bang for the buck. Bottom line. If Via can offerthat to me, then I'll buy it. If Intel offers that, I'll buy that.

And the way things are looking, Intel will have something to offer me in the near future. And that says a lot. An awful lot.

[This message has been edited by Gremsongs (edited 11-19-2001).]

knuttern

11-20-2001, 01:29 AM

Some points:
1. My experience with via isn`t something that I read on the net, it is the results of my own problems.

Case is, with only graphics card installed, I can`t get my soundblaster live to work properly, neither my audiophile 24/96.

2. "prorec.com" or whoever say something therefore, it must be true. Firstly, no lab in the world can produce all variations on setups. Why do you think manufacturers use beta-testers? So far, most "music-pro" opinions have not been technically well-founded. This doesnt mean that they are wrong, just that they may be right for other reasons. And the opinion may be useless in other setups because of that.

3. Out of my experience, I have had 0 problems with my old abith bh6 (bx), and lots with my new abit kt7a (kt133a). Of course, this makes me believe that this mobo is worse than the old one.

The reason I commented on the viatech board is that you will see the same old kt133a problems repeated for newer kt266/266a boards.

Please understand that I don`t want a war on intel vs amd, or via vs nvidia vs ali. It just pisses me off when someone posts "no worries, these problems have now been solved, getyourself a kt133a board and you won`t regret it". Well, I for one regrett it. I have the newest drivers and bioses and 4-iin-ones, but it simply wont work. It`s as easy as that.

Knut

Groovydaz

01-15-2002, 11:06 AM

Firstly there are a number of problems related to the via chipset and if the 'professionals' dont tell you this then they havent a clue what they are talking about, or rather their computer knowledge is less than their musical knowledge.

Problems occur particularly in windows2k, and it is a major problem with a variety of professional hardware including Soundcards and video capture devices. And allthough via have fied some of the problems many still occur. I have had problems with a Yamaha Aureal card (Wouldnt work at all and as i result i no longer use) and a BT878 TV Tuner card, which have been related to the Via system.

Having read the article there are certain things that i would disagree with/have an alternative view on:

PROCESSOR - Although the athlon is an extremely good processor, it is plagued with the problems of the Via chipset on the mainboard. It is however much cheeper and well worth buying, however i cant see a pentium 4 2ghz being a bad choice, if the money is available.

VIA's KT133a - probably the least reliable chipset for DAW

RAM - DDR seems to be working well and is now ready for the consumer market.

HDD - Maxtor more reliable than IBM? Having used both drives over the years, i have known 3 maxtor drives to fail, compared to 0 of IMB, My Current Maxtor 7400rpm also runs at 50 degrees c, compared to IBM's which are not hot to touch. Maxtor offer slightly better prices though.

GRAPHICS - Matrox G450 - an excellent dual display card, but bare in mind in takes up the AGP slot meaning you couldnt use a more powerful video card - a problem if your into gaming or some aspects of video.

As for:

'So, do you buy your products based on forums or based on expertise??'

I think you'll find there is a lot of expertise in these forums, and i trust 50 reviews/comments made by people who have paid money for a card far more than i do some profesional reviewer who has probably used the card for less than a day and gets paid to use it.

Darren.

subhuman

01-15-2002, 11:30 AM

Sometimes I wonder what other people do on their machines if they "have no problem with their via chipset."

I've built maybe 10 machines with VIA, every single one of them has required hours, hours, and HOURS of debugging, technical support, swapping various versions of 4-in-1 drivers (crazy that some people here think it's "good" they release a new version to fix a new bug almost every week?), BIOSs, etc. I know how to build a machine, and have no problems with any of the 50+ Intel machines I've built -- except 2 related to the Abit BE6 motherboard and highpoint (the BE6 V1 was just a flawed design).

Amongst the problems I have had, with KX133, KT133, KT133A, and recently KT266A, are the hideous PCI bandwidth (some VIA lovers will say "but there is some guy in some foreign country who made a patch which helps fix the problem" to which I say "the problem has been there since the FIRST via chipset, I have wasted literally HUNDREDS of hours trying to solve it, to no avail."

So: If your VIA works for you - consider yourself lucky, be happy, and go make music, don't bother standing up for VIA, if their product is good, it will stand up on it's own. I am an experienced builder who sells machines to a very proliferate clientele who are extremely demanding of their work stations (often times syncing it with full quality video), and this PCI problem is a show stopper, quite literally. Maybe next time they'll get it right?

In the meantime, Intel chipsets have been rock solid reliable: 440BX, i815E, i815EP, i845, i845-D, i850 are all excellent for high end music and video work. My current choices are of course i845-D and i850.

Please save your flames for someone else, I have heard them all before. http://www.audioforums.com/forums/wink.gif

Robert D

01-15-2002, 12:22 PM

Sub H - Flame, flame, flame, flame, flame. http://www.audioforums.com/forums/tongue.gif

Still running my 2nd 440BX chipset, still logging hundreds of hours of boring, no crashes, no tweaking, DAW use. Lonely as the Maytag repairman - maybe I'll download a virus so I can have some fun like the VIA users. http://www.audioforums.com/forums/smile.gif On second thought......maybe not.
Cheers, RD

Agnishvatta

01-15-2002, 12:29 PM

Subhuman, although I now refuse to use via I have had goodluck in the past. Not as many people use Creamware products as other products and because of this people will generally have better luck than you who are building Creamware based pcs.

RME has used the Asus A7V as the Athlon referrence motherboard for sometime now and there is a reason for that. It simply works with their products or at least did.

The pci bandwidth may be small and incompatible with many dsp cards, but for a lot of people it works. I used to get 1.5ms latency with many tracks on my old Gigabyte KT133 based pc with an EGOSYS Waveterminal 2496. I have since sold it and the user is happy.

I sold my via board because I was anticipating the arrival of high bandwidth pci devices. I now have an nForce based pc, but I have nothing to test it with yet.

Rv-Sound

01-15-2002, 08:34 PM

My limited experience with Via chipsets is based on my recently aquired new Daw. Its somewhat similar to prorec's RYO 1.4 athlon, but with a pioneer dvd and plextor burner. I wanted to go this way because Its so clear , even when the counterparts are not, that amd's chips are far superior than Intel's now. I'm using win2k SP2 with an aardvark DP 2496. My first experiences with this system were ok. System was working good, stable, but at the end, not as I expected. I was very concern about a PCI bus efficiency on a reading on my DP 2496's control panner showing 47% at all times, no matter what tweak-a-do I did. There was no cracking or pops, but somehow the performance wasn't there in terms of track count, latency and plugin usage. So I did a lil' research on many VIA related websites (mostly viahardware.com). I came to know the PCI bug implementation on via's kt133a and 686 chipsets on certain mobos, causing performance problems with HD's, soundcards and video cards. I also found about a very clever guy wich made a patch (more like a driver) to work along these chipsets to fix the PCI latency problem.
Anyone can find the nfo here: http://www.networking.tzo.com/net/software/
Since I installed his driver, my system has been working damn fast, there was a huge performance bump. My DP 2496 now shows 100-99% PCI bus efficiency at all times, I've got it to work at 2ms latency on some programs using ASIO (something that just didn't happen to me before), track counts also increased a lot more, and also plugin usage. What can I say, the bug ceratinly is there, and I spended to much time tweaking, I would have expected the fix officialy from VIA and not from some other fella. But at the end, I got it to work damn good and stable, wich is what I wanted from the begining, and took a chance to learn more advanced pc stuff, like adjusting the bios, picking the right drivers, etc, etc. So, even with the tweaking, formats, and that, I'm very happy with the result. You can get more nfo on http://www.viahardware.com

subhuman

01-16-2002, 09:14 AM

amd's chips are far superior than Intel's now

That is so last month.. http://www.audioforums.com/forums/smile.gif They are now only superior in terms of price: a P4 2.2Ghz is faster than even an AMD 2000+ http://www.audioforums.com/forums/smile.gif

Rv-Sound

01-16-2002, 12:13 PM

Yup, but you r taking in account just the clock speeds, not the other specs.

subhuman

01-16-2002, 12:21 PM

Actually, I am taking account benchmarks for PCI throughput, CPU, and Memory Bandwidth (chipset related, but you can't buy an Intel chipset for an AMD). Intel is clearly the best at all of them - and includes better CPU core protection (heat spreader), runs cooler (and therefore, quieter), and has better thermal protection (runs even without a heatsync in case one ever comes off).

The main advantage AMD still has, is cost, an AMD 2000+ is quite a bit cheaper than the faster (just a little bit, and not just in clock speed, but still) 2.2Ghz P4. Of course, I can't afford the 2.2Ghz right now either. http://www.audioforums.com/forums/smile.gif

Finally the other specs -
AMD 2000XP cache size - 384k cache
Intel Northwood 2.2Ghz cache - 512k

AMD 2000XP process (lower is bettr) - .18 micron
Intel Northwood 2.2Ghz - .13 micron

AMD 2000XP clock speed - 1.67Ghz
Intel Northwood 2.2Ghz - 2.2Ghz

Chipset related:
Best number of reverbs I can get for AMD (all VIA/AMD76x tested) - 5 reverbs (see other posts for testing methodology)
Intel - 10 reverbs (twice the performance for my purposes)

And now, the part where the Northwood loses
AMD 2000XP price - about $300
Intel Northwood 2.2Ghz - about $650

Finally the difference in speed between these chips is very small. Northwood is only slightly faster, but is in nearly all tests. AMD is the better value. Both CPUs are excellent technology.

[This message has been edited by subhuman (edited 01-16-2002).]

Agnishvatta

01-16-2002, 03:16 PM

Intel is clearly the best at all of them - and includes better CPU core protection (heat spreader), runs cooler (and therefore, quieter), and has better thermal protection (runs even without a heatsync in case one ever comes off).

Subhuman, I find your statements extremely misleading and I'm starting to think that your opinions are biased. You say the P4 "...runs cooler (therefore, quieter)", but something is not quieter because it is cooler. With the right heatsink/fan combo (Alpha 8045/Enermax adjustable fan--the quietest) any type of cpu can run virtually silent. Fans can only get so quiet and fyi I have run an Athlon XP without a heatsink and it simply shuts down. As for your other statements, you say:

AMD 2000XP cache size - 384k cache
Intel Northwood 2.2Ghz cache - 512k

But you don't mention anything about whether it's all level two cache or partly level one as well. The Athlon still has a higher level one cache of 128K compared to intels 20K. Higher cache does not always give better performance anyways.

AMD 2000XP process (lower is bettr) - .18 micron
Intel Northwood 2.2Ghz - .13 micron

Now you mention the micron process, but this is not directly related to performance. It only gives the P4 an advantage when overclocking or for simply bringing out higher clocked cpus. So when comparing what the user will actually experience when using a .18 micron and a .13 micron this becomes a non-issue.

AMD 2000XP clock speed - 1.67Ghz
Intel Northwood 2.2Ghz - 2.2Ghz

Here you would dare to directly compare clockrates and then simply leave it at that. The P4 has an IPC (instructions per clock) of six and the Athlon has an IPC of nine. How about the number of floating point pipelines which is important for audio? For the Athlon it's three and for the P4 it's two.

Chipset related:
Best number of reverbs I can get for AMD (all VIA/AMD76x tested) - 5 reverbs (see other posts for testing methodology)
Intel - 10 reverbs (twice the performance for my purposes)

Here you are talking about chipsets again, but you still haven't provided data on the nForce, SiS735/745, AMD 760MPX or the ALI Magik1 revision C. Even when you do provide that data you will have to use other tools than just Creamware products to get the data. What if these bandwidth issues are solely related to Creamware products? You mentioned in another post that you found a new tool to benchmark bandwidth across the pci bus, are you still going to use it?

And now, the part where the Northwood loses
AMD 2000XP price - about $300
Intel Northwood 2.2Ghz - about $650

Finally the difference in speed between these chips is very small. Northwood is only slightly faster, but is in nearly all tests. AMD is the better value. Both CPUs are excellent technology.

I'm sorry but your respected statements will lead many to make malinformed decisions. As soon as I read what you said above about the P4 being slightly faster in nearly all test I went looking for this evidence. Let me tell you I did NOT find it! Check anandtech.com to see what I mean. The P4 2.2GHz and the Athlon 2000+ exchanged blows and it was as simple as that.

[This message has been edited by Agnishvatta (edited 01-16-2002).]

subhuman

01-17-2002, 10:08 AM

Here check out this link: http://www.simhq.com/simhq3/hardware/features/cpudogfight/index1.shtml

And of course I think the number of choices of chipsets for AMD is nice, but actually a disadvantage since none are up to the quality of Intel chipsets. The gap is closing, and I still haven't tested nForce or the new SiS745 (but the new SiS is basically the same as the old chip with a new memory interface).

Next, as far as Anandtech tests - yes - AMD is faster for certain 3D rendering and AutoCAD tests, Northwood is faster in almost all games except the non-P4 optimized Serious Sam, but we're talking about Content Creation here as audio workstations, and those are the numbers I am refering to!

The ALPHA PAL8045 is a 3rd party heatsync that (personally I love) must be purchased independantly of the CPU. Intel's stock heatsyncs for P4 are huge, and pretty quiet, and they come with the CPU. AMD's stock heatsync is incredibly loud, and you have to chose your motherboard around(!!) your heatsync to get something like that Alpha 8045 to fit. Then you have to source a quiet fan - and a quiet fan will mean the CPU will be running much hotter than I'd like (and hotter than an Intel). I don't know about you, but I have over $4000 in heat generating DSP cards in my computer, so lower CPU heat is really important to me. To get an AMD running as cool as an Intel, you need a 3rd party fan combination that is carefully selected, and that also in turn limits your choice of motherboards. But you are right - there is always some workaround to get around these things.

Finally - yes - I am extremely biased towards the better performing product (for what I do). I get sick of hearing that AMDs are the only thing to get (and I will say it ONCE again - I think AMD CPUs are incredible tech, and a real bargain) -- when people don't understand or consider the full platform -- to me, the chipset is as equally important as the CPU. Let me say that again: *TO ME*, the chipset is equally as important as the CPU. Here, AMD has never really had anything close to the Intel solutions! This is what frusterates me! And I continually get comments from people who really don't use their machines in the same way, saying that "via works fine" - you know what? They are right -- for many people, Via does work fine.

Again - both CPUs are really awesome technology. And I'll leave it at that. http://www.audioforums.com/forums/smile.gif

(Here's to hoping nForce rocks... I have one on order to test. http://www.audioforums.com/forums/smile.gif )

Agnishvatta

01-17-2002, 11:06 AM

http://www.simhq.com/simhq3/hardware/features/cpudogfight/index1.shtml

It is clear that they didn't use the amd patch for sysmark 2001. That benchmark is null, unless you can see the individual scores of each component.

And of course I think the number of choices of chipsets for AMD is nice, but actually a disadvantage since none are up to the quality of Intel chipsets. The gap is closing, and I still haven't tested nForce or the new SiS745 (but the new SiS is basically the same as the old chip with a new memory interface).

The point is not how many chipsets, but that you have not tested all chipsets. Even when you do test them all, you need to test with something other than just Creamware products. Otherwise the data you give may not represent the true picture.

Next, as far as Anandtech tests - yes - AMD is faster for certain 3D rendering and AutoCAD tests, Northwood is faster in almost all games except the non-P4 optimized Serious Sam, but we're talking about Content Creation here as audio workstations, and those are the numbers I am refering to!

Again, you have to look at the individual test in sysmark 2001. Suppose the Athlon XP was faster in every test, but performed poorly on just one (such as mediaplayer 7).

Finally - yes - I am extremely biased towards the better performing product (for what I do).

No, you are biased because you proclaim what is the best before you have tested all of the alternatives. From yourdictionary.com:

Main Entry: 4 bias
Function: transitive verb
Date: circa 1628
Inflected Form(s): bi.ased or bi.assed; bi.as.ing or bi.as.sing
1 : to give a settled and often prejudiced outlook to...

As for the heatsink/fan thing, most people buy Athlons oem, not retail. So maybe for the average consumer who is afraid to fiddle, the P4 wins here.

[This message has been edited by Agnishvatta (edited 01-17-2002).]

subhuman

01-17-2002, 11:17 AM

The only ones I haven't tested are nForce (on it's way), and SiS745 (because I suspect it will be identical to SiS735).

Anyway, have a nice day.

Agnishvatta

01-17-2002, 12:06 PM

What were the results for the ALI Magik1 revision C, AMD 760MPX and the SiS 735?

knuttern

01-17-2002, 12:11 PM

It is disappointing that even in a forum for setting up music systems, the same old intel/amd war repeats itself.

Point #1: Intel and Amd are virtually equall in terms of speed today. Depending on what tests and what time of the month, no cler leader can be chosen. And now I`m talking about testbench programs producing numbers. In real world, there should be even less difference.

For dsp intensive applicaions, the amds might have a lead in non-optimized applications (best fpu), while the p4 might be best for sse2 applicaions (not very many - yet)

point #2: given the number of users out there, you will find stories of failure and success with both intel and via chipsets. But via even admits themselves that they have serious problems with their pci performance. This is an issue that isnt tested in normal gaming/office tests, as it isnt that important for them. Check out the forum at www.viaarena.com (http://www.viaarena.com) . Via regularly ships new 4-in-1s but don`t be fooled: no single set of drivers can be said to fix the problems consistently. Power users have to try perhaps 5 4in1s against 5 mobo bioses against 5 soundcard drivers. This consumes time!

regards
Knut

Oh and sorry that I didnt introduce me properly the first time. I am a 24 year old norwegian part-time musician while doing my masters in signal processing.

Agnishvatta

01-17-2002, 12:34 PM

knuttern, I agree. I am only debating because Subhuman has too much influence to be passing on inadequate information about why the P4 solution (including available chipsets) is superior to the Athlon solution. I actually never even said that the Athlon solution is better (re-read my previous post).

Subhuman, could you tell me more about your test? I need to know how many sharc dsps you need to get 10 reverbs (these are the Timeworks Reverbs right?). I also need to know the brands of non-via based motherboards (including revision #) you used for testing both the P4 and Athlon solutions, as well as the "pci latency value", pci slot(s) used, bios version, ram (amount of sticks, etc...), "hardrive controller", hardrives, Windows OS and any patches that you may or may not have used.

Thanks

subhuman

01-17-2002, 04:57 PM

And again - I think I've said this many times - I like AMD Cpus quite a bit (this isn't what I'm talking about actually), and would use one if they had a platform that had PCI performance that equaled or surpassed that of Intel, without all the extra fiddling around. I *know* you "can" get a working VIA chipset solution, and I know all about patches, H.Oda's tools, etc - but these don't correct problems with the chipsets (back on topic here). I simply don't have the time to mess with trying to get something I can get from an Intel system when I plug it in. (But I have spent the time over and over, with each VIA release, trying many different 4-in-1 drivers, BIOSs, and even did get a SLIGHT performance increase with the VIA chipset by using some of H.Oda's tools. Trust me, I have done my diligence here - have you?)And that is the only point I am trying to make here.

I am not trying to say my tests are scientific - but they are repeatable if you have the hardware. I am also not saying everyone has the same expectations from their hardware - in fact, I have said many times, that a VIA is fine for many users. I do think it's sad that people begin to accept a lower performing solution than what I consider "standard." (440BX/i815/i845/i850 being those PCI standards).

I have also purchased a PCI bandwidth
measuring device, so I might go through all the effort to make quantify this for everyone else, at some point too - but it's already completely clear from the realworld tests which I have described before on this forum. Do I really care if people buy a VIA system? No. http://www.audioforums.com/forums/smile.gif I don't work for Intel, AMD, or VIA. I also don't mind if you get a different brand audio card - and BTW, I have reproduced the problems using echo Layla, midiman Delta 1010, and RME - in fact it's quite easy to do if you have a RAID system or video editting hardware on the PCI bus running at the same time. http://www.audioforums.com/forums/smile.gif

I won't be testing AMD760MPX - you can't use most PCI cards in the 66mhz/64bit slots, leaving you with only 3 usable slots - and because most boards don't have onboard usb (usb bug with the new Northbridge or something) and ship with USB 2.0 PCI cards which take one of those slots, it's simply not a viable solution for what I'm doing. It is a pretty cool platform though (and I think a step in the right direction)! I did test AMD760MP and it had very good performance, in dual CPU mode it equalled the AMD761 single processor in PCI (both tested separately under winXP for my own comparison). And let me say it again, the test is not because anyone needs that many reverbs, but because it's easier than loading a certain number of audio channels onto a fresh machine to test it that way. There is a correlation between the two numbers and I have a formula I use, but since I'm already having to explain so much stuff to you already, and already having to do a bunch of extra work to give you precise, reproducable scientific data to back up my claims, that I am not going to share that until I have the conversion pinned down to an exact number for you.

BTW - it only takes a Luna (3 DSPs) to max out most VIAs. And Luna v3 comes with a free masterVerb (Creamware reverb) which works to show the test.

BTW2 - the PCI problem I am describing is also pretty easy to find out about on the internet, VIA has admitted it, and other tech sites describe it complete with the scientific numbers you seek. A RAID array easily saturates the entire PCI bus on a via chipset, and you get far less throughput using one, than basically any other chipset on the market.

BTW3 - nForce isn't really an option since you're paying for onboard sound video lan, and they won't be used by myself or anyone that I help out. But it will still be tested in time.

Anyway ... I am no longer looking for new computer customers, I send everyone to dawbox.com soundchaser.com or liquiddaw.com - and let them buy whatever hardware those guys are selling. And last I looked at least 2 of those places were selling VIA machines and both didn't understand the issue when I tried to bring it up (and both said I didn't know what I was talking about, but I forgive 'em)

...I think it's simply easier to let people struggle than to try to explain why they should go down a certain path. Perhaps that's why other DAW stores just sell what people want instead of what performs best by careful testing.

subhuman

01-17-2002, 05:05 PM

Let me say it again:

The AMD system is cheaper (I have always said it was a better value than any Intel), and when paired with an AMD761 chipset, you can adequate PCI performance for most tasks.

Re read my posts. I'm not saying everyone should buy a P4 - I am just telling you why I bought one. I have also offered AMD solutions in the past - and still recommend the AMD761 or AMD760MP until something better comes along (yes more testing is needed on the chipsets that I would never consider for my own desktop - Ali, SiS, nForce, AMD760MPX).

This isn't AMD vs. Intel (CPUs), it's a lot more like VIA vs. Intel (Chipsets).

Agnishvatta

01-17-2002, 05:48 PM

I have also purchased a PCI bandwidth measuring device, so I might go through all the effort to make quantify this for everyone else, at some point too - but it's already completely clear from the realworld tests which I have described before on this forum.

I'm eager to see these new benchmarks. Also, where can I get one of these PCI devices?

And let me say it again, the test is not because anyone needs that many reverbs, but because it's easier than loading a certain number of audio channels onto a fresh machine to test it that way. There is a correlation between the two numbers and I have a formula I use, but since I'm already having to explain so much stuff to you already, and already having to do a bunch of extra work to give you precise, reproducable scientific data to back up my claims, that I am not going to share that until I have the conversion pinned down to an exact number for you.

Until you are ready to share, your claims should be taken with a grain of salt then. Waiting on the conversion formula is understandable.

BTW - it only takes a Luna (3 DSPs) to max out most VIAs. And Luna v3 comes with a free masterVerb (Creamware reverb) which works to show the test.

Okay, this is getting close to what I need to know. So when you performed the test you used a Luna with the masterVerb? And, with the AMD 761 you could only get 5 masterVerbs? With the Intel solution you can get 10 masterVerbs using the Luna?

BTW2 - the PCI problem I am describing is also pretty easy to find out about on the internet, VIA has admitted it, and other tech sites describe it complete with the scientific numbers you seek. A RAID array easily saturates the entire PCI bus on a via chipset, and you get far less throughput using one, than basically any other chipset on the market.

VIA is the only one though. It was pretty clear to me that you were referring to the AMD platform in general when you said,

Actually, I am taking account benchmarks for PCI throughput, CPU, and Memory Bandwidth (chipset related, but you can't buy an Intel chipset for an AMD). Intel is clearly the best at all of them - and includes better CPU core protection (heat spreader), runs cooler (and therefore, quieter), and has better thermal protection (runs even without a heatsync in case one ever comes off).

BTW3 - nForce isn't really an option since you're paying for onboard sound video lan, and they won't be used by myself or anyone that I help out. But it will still be tested in time.

It's pretty odd to hear you say this. In your previous post you were usually pretty enthusiastic about nForce. Actually, a lot of people would prefer the all in solution. The onboard video doesn't degrade ram performance much at all either. Of course for dual or quad monitor solutions the onboard video is a waste. As for the lan it is definately good, although if it were gigibit ethernet I would be happier. The onboard sound has it's uses too (I'm not talking about games).

...when I tried to bring it up (and both said I didn't know what I was talking about, but I forgive 'em)

This is why you should provide the whole story and is why I asked for the detailed information.

...I think it's simply easier to let people struggle than to try to explain why they should go down a certain path. Perhaps that's why other DAW stores just sell what people want instead of what performs best by careful testing.

Well, if you carefully wrote up a nice piece on this whole bandwidth issue, with most of the details, then people like me wouldn't ask many questions. Sorry if it's such pain, but if you could give me the details then I would surely understand it and definately appreciate it.

(yes more testing is needed on the chipsets that I would never consider for my own desktop - Ali, SiS, nForce, AMD760MPX)

Alright, but why would you never consider Ali or nForce? As I'm sure you know, there are a few new nForce chipsets on the horizon. One of them the 415 has no onboard video and supposedly boards might appear as cheap as $100.

This isn't AMD vs. Intel (CPUs), it's a lot more like VIA vs. Intel (Chipsets).

Alrighty then....

[This message has been edited by Agnishvatta (edited 01-17-2002).]

Alexis

01-20-2002, 09:05 AM

Originally posted by Agnishvatta:
Of course for dual... monitor solutions the onboard video is a waste.
On Tom's mobo/forum I read that it is possible with a non AGP (PCI) video card.
Besides Even if a "lite" 415 version is shipped soon I'd personnaly go for the full 420. My reasons: IRQ free, CPU load, compatibility, possibility of dual monitoring (if true) and anyway better for moderate gaming than any Matrox card.
In fact I'm wating its PCI bandwidth results (important for me as my main audio card is DSP Factory) to buy it.

[This message has been edited by Alexis (edited 01-20-2002).]

Alexis

01-20-2002, 09:34 AM

Very informative your dialog gentlemen (and maybe ladies?). Ones again that's a great forum.
Subhuman,
When you're talking about reverbs, it is understood onboard ones, not plug-ins? Or plug-ins are they dependable on PCI bandwidth, (and not only CPU power) if they are performed by a PCI DSP based audio card?

Andrei

02-03-2002, 01:26 PM

Sorry if my reply is too late, but all
people in this thread somewhat are right
Most of horrible VIA issues symply don't
exist. At least, starting from KT133A.
Most of problems are because of buggy drivers
and probably manufacturers haven't test them
enough. Other half lf problems related to
typical windows issues.
Mr. Xanto notice that his secret is dedicated
system (no net) and it is a point. It's
no way really monitoring what is going
inside of the box.Benchmarks (specially
from consumers websites) have nothing to do with reality and what actually new we can expect?
For other hand no doubdt that intel chipsets perform better - but how critical
is this difference?
One example 2comps running the same CPU
intelPIII and AMD. Sound card MarianMarc2
with Driver1.02.AMD is getting popups(plugins are the same) in b.s. 72, Intel
can show 64. In some more hard teste Intel
superiority might be bigger, but it doesnn't
mean, that you must get Intel AMD might somewhat is winning in completely native
enviroment( I mean CPU utilisation).
So it's better to purchase cards wich are able to work fine with both chips. Somehow
I believe that if your card can't work with
AMD it also will never be the best performer with Intel
But once more Intel chips are better and what Knut said about PCI management is reliable information . But the most important is quality of drivers and optimisation of OS Even the best driver
need priority to show it's abilities.
I personally never use 4in1 driver
(MSI KT7proA with VIA133A) - no need,
All my ptoblems disappear generally, when one day I remove IE. Ihave been using a lot
SBlive! with APSdrivers, ofcource, and it was quite OK for this pricetag card.
Regards.

xanto

02-03-2002, 09:08 PM

Hello!

From all you said what I really agree with is that Yes, it´s too late to replay to this forum...

wish a nice day!

xanto

Alexis

02-11-2002, 07:22 PM

Originally posted by Agnishvatta:
I'm eager to see these new benchmarks. Also, where can I get one of these PCI devices?

Agnishvatta,
Could you finally get this precious tool to make some scientific benchmarking of your nForce? You are right. After hearing everyone's opinion, the best thing will ever remain, some solid mathematical figures.

Agnishvatta

02-11-2002, 07:52 PM

No, I Subhuman never did tell me what that device is. If he does decide to tell me and if it seems good then I'll definitely get you some feedback. When I have more time I'll do some pci bandwidth test using raid cards, uad-1, gigibit ethernet and several different types of high bandwidth audio devices (like a couple of LynxTwos streaming a bunch of channels at 192kHz, 32/24bit).

I'm going to hold off on Creamware stuff for now. I think that the bandwidth problems may be isolated to their dsp cards. Because even after the pci bandwidth has been increased with patches for the via chipsets, there is still a major problem. I think it's something other than just bandwidth relating to poor performance when using Creamware products with VIA, SiS, etc....

When it comes down to it, plenty of people are using UAD-1 and Powercore cards along with raid and other audio devices without any problems.

A little over a year ago RME was recommending the Asus A7V. Tons of people had problems using that board with other audiocards. Point being, that not all cards will perform the same with every system and yes this would relate to the drivers as well.

Btw, Alexis, did you ever get the nForce? Supposedly Creamware stuff isn't working well there either. But people are using cards like the UAD-1 with no problems.

Alexis

02-13-2002, 04:49 PM

I've got a DSP based audio card, and a very capricious one :-) (DSP Factory).
When (!) it works, I'm fully satisfied by the performance, though meanwhile am frightened to tuch or change anything on PCI slots. Here my concern about nForce's PCI behaviour. All my DAW becomes useless if it won't work properly and acqisition of high end Athlon CPU will be a simple waist of money. The chipset itself suits 95% my needs (dual screen option would bring it to 100%). I love the idea of full integration, LAN etc. Just fed up with trillion drivers installs, compatibility and so on. Speed also is good. I'll be really desapointed if nForce turns out to be a poor PCI performer. My 950MHz Celeron permits me to do my current work, though for my future project I will need a more powerful system. Subhuman promised us some scientific testing with his PCI tool (in fact, as you say, the Creamware gear testing has only a relative interest for me), and as soon as the verdict will be here I am going to upgrade. But here again, the results should be highly explicit for me to avoid nForce :-).

bassdude

02-17-2002, 04:36 PM

I believe the UAD and Powercore cards have onboard memory. This is probably why they suffer less PCI throughput problems http://www.audioforums.com/forums/smile.gif

Valentin

02-17-2002, 07:52 PM

I keep hearing about the pci problem with the AMD chipsets. So what is it, all AMD chipsets or Via where the problem is? I thought it was only the Via set that had it as the tests here suggest http://www.tecchannel.de/hardware/817/index.html
What then is the best chipset to get for an AMD cause theres no way i can (will) afford an intel. The Ali and SiS sets seem to do ok, not quite as good as the intel but certaintly not so poor as to be a no-go area!

Secondly, when does the pci issue become an issue? If im monotoring multiple inputs and outputs? If im using many tracks in a sequencer? Is it moreso if i move upto 24/96? If im recording only?
Is the pci bandwidth exactly the same as in intel chipset other than the max limits or will a via cause clicks long before its limit is reached because of the stream constantly being interrupted by other devices needlessly claiming the bus?