The No.1 Website for Pro Audio
 Search This Thread  Search This Forum  Search Reviews  Search Gear Database  Search Gear for sale  Search Gearslutz Go Advanced
Has anyone here built a Xeon Dual Quad yet ? Audio Interfaces
Old 29th August 2007
  #1
Lives for gear
 
True North's Avatar
 

Has anyone here built a Xeon Dual Quad yet ?

I know 'Ten' built one a little while ago , just wondering if anyone else has.

Motherboard and specs would be nice - anyone here built one ?

Thanks
Old 29th August 2007
  #2
Lives for gear
 

HI,
if you have UADs (more than 1) forget it!
and now with Steiny saying it will NOT have a fix for the 8 cores for some time now i would not bother period.
better to wait for AMD Barcelona or build a single quad core.

Scott
ADK
Old 29th August 2007
  #3
Lives for gear
 
crypticglobe's Avatar
Quote:
Originally Posted by jcschild View Post
HI,
if you have UADs (more than 1) forget it!
and now with Steiny saying it will NOT have a fix for the 8 cores for some time now i would not bother period.
better to wait for AMD Barcelona or build a single quad core.

Scott
ADK

Hey Scott. What are you talking about with UAD-1 cards? Nothing a magma (pci to pci or PCIe to PCIe won't solve).

Built a couple here. No real problems other than the fact that you just don't get all the power you would expect to get with Steinberg applications. A fix is being worked on though...
Old 29th August 2007
  #4
Lives for gear
 

Hey Steve,
even with Magmas (why should you have to spend that kind of money)
its still hit and miss.

there is absolutely no point to a Dual Xeon (or dual quad) and Steiny products.

a single quad outperforms the Dual dual Xeons
the dual quads fall on thier face with steiny..
making the single Quad more powerful.

i can run 4UADs and a PCI or Firewire interface in a single quad without having to pay for a magma and get better performance for less $.

its rather simple to me.

this is one time where the "Intel just works" mantra falls flat on its face.

i would rather sell a Opteron even if its less powerful.

with Barcelona just around the corner i cant wait.

Scott
ADK
Old 30th August 2007
  #5
Lives for gear
 
True North's Avatar
 

Oh boy, I missed all the hoopala on my very own thread... damn

Scott - on your site the Dual Quad Xeons seems to work pretty well on the Magneto test VS the Single Quad -at 64 sample latencies performance appears to be almost identical to a single Quad - I don't need to track at lower than 64 sample latencies.

I gathered the following performance increase percentages off your benchmarks
1024 samples 39%
512 samples 33%
256 samples 22%
128 samples 13%

I guess if you are spending more than double on Xeon Quads you would expect more or close twice the performance - so it is not a very cost effective solution.

Are you expecting that the AMD Barcelona Dual Quads will show a noticeable performance boost with Steiny products?
Old 30th August 2007
  #6
Lives for gear
 

Hey North,

the benchmarks on my site are a simple test designed to show power.
and done with the RME fireface 800

in actual real studio use (read :those with UADS/Pocos) or even say 2 rme madi cards with all 128 i/o active.
the picture changes rapidly.

someone with 2 rme and tracking @ 1024 its fine. (more power than they need really)

however start doing overdubs @ 128 buffer and good luck with that.

At this point i only like selling Xeons to someone who does not own Steiny.

now whats odd ( but i know why) with video editing (particularly render times or 3D animation)
the Dual Quad Xeons slay anything else by a huge margin.

as to the Barcelona i dont nessearily expect much better performance for steiny ( i do expect some)
but an overall better system stability with UAD/Poco etc in a heavy studio use.


Scott
Old 30th August 2007
  #7
Lives for gear
 
crypticglobe's Avatar
I agree with all your specs Scott. I guess I am just thinking ahead. Dual Quad is going to be freaking AMAZING for Steinberg products fairly soon. They are just working out their issues.



p.s. Why a magma? Well... you know I still have a line on very, very affordable ones. Pretty much close to an unlimited supply. And for those that already have a bunch of the PCI cards..it's cheaper than re-purchasing the PCIe versions.

A single Quad is definitely the best bargain right now for sure. But when Steinberg gets their act together... Dual Quad is going to to be totally rocking for sheer horsepower.

I am excited about Barcelona too!! I still haven't been able to glean any real clarity about whether or not that they will be available at full performance for Socket "F" like they originally said it would be or not though.

Any clues?
Old 30th August 2007
  #8
Lives for gear
 
True North's Avatar
 

Hey Steve and Scott,

Thanks for the replies. I do use Cubase SX3 (I also own Reaper and PT) and I currently own a 7 slot Magma chassis for my 4 UAD Cards. I know a lot of post guys use a hi # of I/O's but I only record and mix music.

I did want to address some of Scott's points a little further

Quote:
Originally Posted by jcschild View Post
in actual real studio use (read :those with UADS/Pocos) or even say 2 rme madi cards with all 128 i/o active.
the picture changes rapidly.

someone with 2 rme and tracking @ 1024 its fine. (more power than they need really)

however start doing overdubs @ 128 buffer and good luck with that.
I have 2 Lynx AES16 Cards in my current system and I use software monitoring so I of course need a 128 buffer minimum for recording. I currently run a Dual Dual Opteron 280 rig and I can run plugs while recording 20+ tracks simulataneously.

Is a Dual Quad Xeon going to run worse than my Dual Dual Core (DDC) at a 128 buffer. It would seem that playback performance @128 on a Dual Quad is substantially better than a Dual dual core opteron so it would stand to reason that it would be better while recording as well.

Is my reasoning accurate or does the performance on a Dual Quad while recording at a 128 buffer, actually get worse than a Dual Dual Core Opteron rig ?

Quote:
Originally Posted by jcschild View Post
At this point i only like selling Xeons to someone who does not own Steiny.
Is this mainly because the performance improvement to price ratio is bad in your opinion.

Quote:
Originally Posted by jcschild View Post
as to the Barcelona i dont nessearily expect much better performance for steiny ( i do expect some)
but an overall better system stability with UAD/Poco etc in a heavy studio use.
ummmm, What is a POCO ?

Have there been stability issues in running a dual quad with UAD's in a Magma ? If so could you please expand a little, thanks

Aside from being an impatient power hungry freak for technology I am also in agreement with Steve as to the longer term viability of Dual Quads eventually being supported by Steiny. Steve is obviously more priveldged with the level of information he is receiving than I, so if he has hope, I trust that there must be a reason for that. heh

Steve,

Does it kill you somtimes having to hold back on information or is it easy for you? I would love to know the inside scoop as well but I think I would go nuts having to keep it to myself all the time.

Thanks again


P.S. - I wish there was more geek talk on this site. Right now the Nuendo forum seem to be the only place where guys are talking intelligently about bleeding edge technology, and I own Cubase so I can't post ........ oh well
Old 30th August 2007
  #9
Lives for gear
 

Hey Guys,

well i might have agreed to the "forward thinking" idea last week.
after seeing what Fredo posted forget it. now weare talking 6-12 months before its fixed the right way.

Screw that.....

a single quad core will outperform your opterons..... and the dual quad Xeon at lower latency

so again Xeon is crap for steiny products....

i could post a lenghty email as to why it happens

there is alot more going on than lousy stieny coding
there is PCI to PCIe bridge issues, Dual FSB is another issue screwing with system timing. unstable PICe clock and several other things.

the subsystems themselves are not well designed for real system time performance. (which is what audio is at low latency vs video)

after having an Intel engineer here last time (he comes every quarter)
we discussed our hypothesis with him and he agreed and was to some degree aware of some of the issues.
his comment: the next chipset release will be better.

once Intel finally releases CSI (their answer for Hyper-transport)(2008-2009)
the story will change.

my only hope is barcelona will remove any doubts.
AMD is sending me some in the next week or so.

to make a quick compare (something i know Steve wil understand)

i look at the present Xeon boards and compare them to the Tyan 2885
with all the odd issues and UAD funkyness...

Scott
Old 30th August 2007
  #10
Lives for gear
 
crypticglobe's Avatar
Quote:
Originally Posted by jcschild View Post
Hey Guys,

well i might have agreed to the "forward thinking" idea last week.
after seeing what Fredo posted forget it. now weare talking 6-12 months before its fixed the right way.

Screw that.....

a single quad core will outperform your opterons..... and the dual quad Xeon at lower latency

so again Xeon is crap for steiny products....

i could post a lenghty email as to why it happens

there is alot more going on than lousy stieny coding
there is PCI to PCIe bridge issues, Dual FSB is another issue screwing with system timing. unstable PICe clock and several other things.

the subsystems themselves are not well designed for real system time performance. (which is what audio is at low latency vs video)

after having an Intel engineer here last time (he comes every quarter)
we discussed our hypothesis with him and he agreed and was to some degree aware of some of the issues.
his comment: the next chipset release will be better.

once Intel finally releases CSI (their answer for Hyper-transport)(2008-2009)
the story will change.

my only hope is barcelona will remove any doubts.
AMD is sending me some in the next week or so.

to make a quick compare (something i know Steve wil understand)

i look at the present Xeon boards and compare them to the Tyan 2885
with all the odd issues and UAD funkyness...

Scott
Thanks Scott. There is a XEON motherboard that I am using that is every bit as reliable as my trusty ole S2895's, and as the 2915 socket "F" board that I really like as well. Call me if you wanna know which one it is... lol.

That's cool that you are getting some barcelonas early. Give me a buzz and let me know what you think of them. Most Def very interested.

True North: Yeah... it can be a drag to know some of what's coming down the pike and not be able to talk about it... but I just call up some of the other people I know that have NDA's and we commiserate... lol..

I really think N4 is going to be really amazing and make a lot of people happy.
Old 30th August 2007
  #11
Lives for gear
 

Hey Steve,

I have tried both tyan boards 4 super micro and 2 intel

One works with some things the other with others
Nothing yet is stellar across the board (pun intended)

1 i can get working with UADs (3) and either a FF 800 or an HDSP card.
however toss a Madi in there and it all goes to poop.
(tried PCI and PCIe madi)
every combination of PCI/PCIe UADs.

for most this would be fine i guess, then another wil work ok with a fw MADI but even 1 uad is funky.

Scott
Old 30th August 2007
  #12
Lives for gear
 

Quote:
Originally Posted by crypticglobe View Post
I really think N4 is going to be really amazing and make a lot of people happy.
the only thing that will make me happy is they fix the multicore bug.
i dont give a rats butt about more bloatware features.

and based on what Fredo posted it aint fixed and aint gonna be anytime soon.

soo.... there goes another 100+ users off to another program and several 100's more pissed off and stuck in 2005.

Scott
Old 30th August 2007
  #13
Lives for gear
 
crypticglobe's Avatar
Quote:
Originally Posted by jcschild View Post
the only thing that will make me happy is they fix the multicore bug.
i dont give a rats butt about more bloatware features.

and based on what Fredo posted it aint fixed and aint gonna be anytime soon.

soo.... there goes another 100+ users off to another program and several 100's more pissed off and stuck in 2005.

Scott
Hmmmm...

Well... I can say that none of the new features are "bloatware". Most are things we have wanted for a LONG time, and the rest are huge steps in productivity for post production workers.

As to the multi-cpu scaling issue. All I can say is that sometimes Steinberg takes a while to fix big problems like this... but when they do... they usually take it about 5 years foward... instead of just "catching up". In other words, they try to not only fix the problem, but anticipate how it might need adjustments in the future.

Granted, one would think they would have done that when we had Quad-Core issues.... but I think this time around we will have it.

JMTC.... of course.
Old 31st August 2007
  #14
Lives for gear
 
TAFKAT's Avatar
 

Cool

Quote:
Originally Posted by jcschild View Post
the only thing that will make me happy is they fix the multicore bug.
i dont give a rats butt about more bloatware features.

and based on what Fredo posted it aint fixed and aint gonna be anytime soon.

soo.... there goes another 100+ users off to another program and several 100's more pissed off and stuck in 2005.

Scott

Hey Scott,

I hear ya , and couldn't agree more..

I was wondering when you were going to join in the chorus about Steni dropping the ball on the X-scaling..

I'm with you on some of the other points you raised , but I am not ringing the death bell quite as loudly on the Dual Xeon Platform . True, I currently have a Dual Quad 5355 dev system waiting on some software to keep up at the lower latencies , and like you, prospective clients who were circling waiting for N4.1..

Now that the geniuses have admitted dropping the ball by not prioritising the x-scaling enough to have addressed it 9 months after it was bought to their attention , and with no real ETA on when they will or even if they can fix it - serialisation of the ASIO driver isn't going to be fixed by a hack and patch - the Dual Quads are not on the top of the most wanted list.. :-(

I currently have 5 members with Dual Quads over at DAWbench , all of us Steini users , and I can tell you this is pretty much the last straw.

What pisses me off even further is that even if they do manage to address the scaling issue, there is another inherent issue within Nuendo's resource allocation that tanks after you have loaded a certain number of plugins where there are no resources left to assign to the GUI of the plugins, which then manifests itself into something far more sinister where the whole windows outer skin is effected, and the system simply runs out of resources to do anything past hitting stop.

This is not memory or CPU related, its simply the coding within Nuendo failing to reserve and assign the appropriate resources internally. I reported this 12 months ago , its been confirmed by quite a few others who have pushed the systems to breaking running my benches on the Quadcore/ Dual Xeons.

I'll bet my bottom dollar that the issue is still in N4.1

So I can just see it now, we will get plenty of ventilation about the new features , drag and drop inserts- wow like thats something that took an extra 12 months to code, and the now already overhyped automation that everyone else has had for years, and what ever else they have polished into the mix, but with the core fundemantal premise of the native scalability missing , then its just polishing a turd.. IMO !

SONAR , Reaper and even PT 7.3.2 are scaling correctly to 8 Cores , SONAR even using ASIO drivers, so this whole thing about serialisation of the ASIO driver with the Stein apps is only specific to their implementation - rather sad when its their spec.

Its all well and good to try and remain optimistic , but we have all been dealing with this company at the corporate level for a long time , and if past history is anything to go by, bets are on being more misty optic than optimistic.. LOL

P.S: Ahhh, The Barcelona's, the promised second coming of AMD , I am already feeling the rumblings for the return of the AMDian lobby .. ROTFLMFAO..

Peace

V:
Old 31st August 2007
  #15
Lives for gear
 

Hey Vin,
yeah i am not to happy with steiny...
what i consider the best software and an answer to Pro Fools has me now scratching my head.

Pro fools (normally well behind the curve on tech)
both HD and LE work wonderfully with 8 cores.

Sonar near perfect (1 cpu is always higher)
Reaper perfect.
Logic does it well (on a MAC of course)

as for AMD
like i said i am not expecting a miricle just a catch up. (barely at best)
and expect better stability for mad use systems than what we have now with Xeon..
i know AMD wont touch the Intel on desktop!
Post Reply

Welcome to the Gearslutz Pro Audio Community!

Registration benefits include:
  • The ability to reply to and create new discussions
  • Access to members-only giveaways & competitions
  • Interact with VIP industry experts in our guest Q&As
  • Access to members-only sub forum discussions
  • Access to members-only Chat Room
  • Get INSTANT ACCESS to the world's best private pro audio Classifieds for only USD $20/year
  • Promote your eBay auctions and Reverb.com listings for free
  • Remove this message!
You need an account to post a reply. Create a username and password below and an account will be created and your post entered.


 
 
Slide to join now Processing…
Thread Tools
Search this Thread
Search this Thread:

Advanced Search
Similar Threads
Thread
Thread Starter / Forum
Replies
Tetness / Product Alerts older than 2 months
19
jcschild / Music Computers
4
Seti808 / Gearslutz Secondhand Gear Classifieds
0

Forum Jump
Forum Jump