Hi folks,
Seeing as more and more people, following in OPUS's footsteps, are selecting the Asus - Northwood combo for their new DAWs, I suggest we form Asus-Northwood User Society (ANUS for short) :)
audiohead
Comments
I've also just sent off my order for a brand new P4 Northwood 1.
I've also just sent off my order for a brand new P4 Northwood 1.6a and ASUS (P4T-E) mobo and will be building a DAW for my home studio. Yeah, I know all about the 100MB/s PCI limit bug in the i850 chipset, but I really don't think it'll cause me any problems since I'm not going to be multitracking a bazillion tracks with Ultra160 scsi drives... just a few tracks with lowly ATA/100 drives which really only sustain in the neighborhood of 40MB/s tops. The extra memory bus bandwidth was what I was primarily after.
Will I try to overclock it? You betcha! :p
--Nick
Here's something to make sure you get...the Malaysian P4's...for
Here's something to make sure you get...the Malaysian P4's...for some reason they are safer to OC than any other P4 made elsewhere.
For example...our good new member Jetoney has built a P4 1.6a(see need all hardware and software Guru's post)and he can not OC it unless he raises the voltage core to the chip..now, meanwhile I've got mine going just fine without having to adjust nothing in the BIOS whatsoever! So...hunt em down folks..Malaysian P4's are the way to go!!!
Let us know how your systems spec out guys!!
All hail the ANUS club!! lmao!! what a great name!!
Opus
Thanks Opus for your enthusiastic posts which helped me make up
Thanks Opus for your enthusiastic posts which helped me make up my mind. Indeed, the system was a breeze to set up - no problems (well, actually, I had one - my SBLive wouldn't play wav files in Win98SE, but it was easily cured with the latest drivers update). I haven't tried to o/c it yet since I have some reservations about the effect of PCI bus frequency increase on my main audio card (Audiophile). It is interesting, what audio cards are you guys using and how do they behave when the FSB frequency is set to 133 MHz?
Cheers :)
OPUS, My ANUS machine is up and running. Wow! can't believe
OPUS,
My ANUS machine is up and running. Wow! can't believe the performance. I'm hangin' out at 1.6GHz for a few days to make sure there are no factory-direct problems. Thanks OPUS for being the guinne pig for all of us lowly daw users - YOU'RE THE MAN!
issues:
1) The P4B mobo only accepts 1.5v video cards. I had to put in a really ancient PCI video card (I don't even know the manufacturer or model, but win2k recognized it). The machine did lock up a couple of times during instal "something about memory dump." I suspect the old card may have had something to do with it. Any suggestions on video cards? All of my computer buddies say that nvidia is the bomb!
2) Also unfortunately, I did get audio included on the mo-board by accident. I dissabled it in BIOS but I suppose there's a possibility of conflict with my Layla 24 and Mia 24. Specifically, I dissabled the "on-board audio" but had to leave the "AC97 audio" activated in order for the soundcards to work. the manual said that PCI slots 2 and 6 are "open" as opposed to "shared" so that's where I put them. Any suggestions?
3) I remember you posting some win2k optimization tips, but I can't seem to find the post. could you please give me a link?
Rok on,
speedracer
Yes..most of the newer boards today take advantage of the 1.5V A
Yes..most of the newer boards today take advantage of the 1.5V AGP cards and need that to run...I reccomend any video card except for Svirge based chipsets..
As long as the Vid card is 32MB and lower it should be fine...ATI, Matrox or Nvidia are all fine cards to get...
I have the onboard audio and keep it running with no problems..I have that set as my default audio out for windows..also I have disabled all the USB ports except the USB 2.0 which I have currently running my US428..dont know if there's any difference in performance but why the hell not ..right?!!
If the IRQ's arent sharing with anything then I'm sure the cards are in the correct slots!
http://www.opusaudioprojects.net/Articles.htm
there is the link to the Win2k Optimization doc..altho I will be adding the XP doc within a few days..I'm also going to update the 2K guide as well..I will let you know when it is done..
The XP doc is almost done I just have to finalize it and add in all the fun tweaks at the end!!
Glad to have been the guinea pig for you guys!
Opus :D
just got my pile of parts delivered today from googlegear and wi
just got my pile of parts delivered today from googlegear and will be assembling them as soon as I get over this %@^#$!& strep throat :mad: and my brain isn't frying so much with fever. On a more cheerful note, I did luck out and get a "Malaysia" P4 1.6a chip so hopefully it'll be a good OC'er. I'll post a report when I get the thing built.
Hi fellows ANUSes! Am I glad I just found this forum! I just b
Hi fellows ANUSes! Am I glad I just found this forum! I just bought a p4b266 and a 1.6a GHz two weeks ago and have been pulling my hair out ever since. First my Maxtor HD was bad, had to send it back.... then my floppy was bad - imagine that! ... then I installed Partition Magic, did a few operations fine, then the next day I resized to boot partition, and it crashed big time. Had to up the ante-up for the lastest 7.1 version just to talk with tech support and get it back on line! anyway, I'm up and running now but haven't bought my new daw equip - will do that in a couple of months as $ permit. I'll keep monitoring you folks and see what's up, and if things keep breaking, I would like a shoulder to cry on! Thanks for being here! JT
PS I was told all the lastest video cards are 1.5 volt, the old ones are different. I got a ATI VE dual screen card for $45, as they are discontinued but highly recommended. It's running without a hitch.
I'm pleased that my first post in this forum is join with my fel
I'm pleased that my first post in this forum is join with my fellow ANUSes. Just yesterday I ordered a P4B266C and a 1.6A GHz. Along with two new Seagate drives, 512 MB Crucial, Enermax 350W power supply, and a Matrox G450. I have high hopes of a great system when this overhaul is completed. I'm leaning toward a Delta 66 to finish it off.
Hey, how do you guys feel about DDR ram vs. Rambus for P4? (i845
Hey, how do you guys feel about DDR ram vs. Rambus for P4? (i845 versus i850 chipsets)
6 months ago when I was researching DAWs, the P4 was considered 'not good for audio'... It seemed like the only way to go was AMD or a PIII. Part of the trouble was supposed to be Rambus RAM which was the only choice... something about bad latency. (So at the time I put together an Athlon 1.4 with DDR and haven't got it to work for more than 2 hours without a system crash!! Tried about everything short of buying new mobo. Talk about a huge waste of time!)
Anyhow, I'm looking to build again, and it seems that the whole P4/Rambus is now a non-issue? What happened? Why the change of heart? Now rambus=good?
So I figure that while I could 'theoretically' get an XP 2100 to work, well... now that P4 is apparently resurrected, I'm thinking it would likely be a surer thing. And while the AMD 133mhz system bus should theoretically offer better PCI throughput, my instict is that Intel at 100mhz is probably gonna beat it 'in the field'.
So what's the ideal P4 system look like?
thanks all,
theDuDe
Ideal set up is P4 1.x or 2.x(your choice!) Asus P4B266 mainboa
Ideal set up is P4 1.x or 2.x(your choice!)
Asus P4B266 mainboard with the Intel845D chipset(D standing for DDR support)
The beginning days of P4 basically no software or OS's were truly optimized for their FPU's..now it's okely dokely..especially with XP.
the Intel 850 chipset has a data corruption bug on files 90MB and larger...one may not see this happen to them but why take the chance.
Rdram has never been an issue..it's been the chipset/processor.
As you see here there are a bunch of people duplicating my setup because it works and works like nothin I've seen! fast, furious and stable as all hell!!
Opus
I finally assembled my pile of new parts (P4T-E 1.6a, rambus) la
I finally assembled my pile of new parts (P4T-E 1.6a, rambus) last night and begun playing with the overclocking.
I ended up getting a P4T-E with the CYP clock generator chips instead of the more desireable and now extremely hard-to-find ICS chips so my mobo will never be able to do 533 memory speeds :( , but still at 3X memory bus multiplier, runs at 399MHz memory clock speed. I wonder how 3X at 399 memory bus will compare in thruput to 4x at lower FSB/overall lower cpu speeds? At FSB set to 133 where the PCI and AGP dividers are set to run at normal speed, the machine is rock solid stable at 2.13 GHz with cpu voltage set at default 1.50v.
I'll finish installing the O/S this weekend and then run some benchmarks to compare normal clocking vs. various overclocking speed settings.
I'm kinda wishing that now I would have bought the P4B266 DDR combo instead, but I'll probably be happy enough with what I've got.
Originally posted by Nick Driver: I'm kinda wishing that now I
Originally posted by Nick Driver:
I'm kinda wishing that now I would have bought the P4B266 DDR combo instead, but I'll probably be happy enough with what I've got.
Ya..well I kinda wished I waited for the new intel chips coming out with a norm of 533 FSB!! DOH!!!!!
Opus
Originally posted by Opus2000: Ya..well I kinda wished I waited
Originally posted by Opus2000:
Ya..well I kinda wished I waited for the new intel chips coming out with a norm of 533 FSB!! DOH!!!!!
Opus
I'd thought about waiting for that too, but I'm just too impatient... gotta have it right now so like the old saying goes: ya pays yer money and ya takes yer chances ;)
Still this new machine, even running at stock clock speed, is literally more than twice as fast (feels like 4 times faster) than my old one so I've got no real reason to gripe...especially since the total cost of all these upgrade parts is only about half of what the old machine cost to build three years ago and now I've finally collected enough parts to build a couple complete extra machines, except for monitors, out of all my old stuff so I can finally have dedicated MIDI hosts for my old SB-AWE32 and Gravis Ultrasound cards to hook up to my vintage Ensoniq VFX/SD and Seqential Circuits Prophet-5 keyboards.
Yup...impatient I am as well! I'm surprised I didnt get one soon
Yup...impatient I am as well! I'm surprised I didnt get one sooner but I knew waiting a little bit was ok...
nice to go from a slow machine to a super duper machine aint it? I went from a PIII650 to the P4 and my jaw drops everytime I use it still! lol!
Yeah, I've got extra parts lying around as well for a couple of machines too...
just need some hard drives and a couple of floppy drives and I got my self one extra machine for the VST system link(going to run Cubase 5.1 on it) and another machine to setup as a server/firewall system to feed internet to the main internet/graphics machine..
I figure my roomates computer(the graphics machine) is kind of bogged down with Norton's and all the internet stuff so if I putthat stuff on the server he'll get a boost in performance for his 3D animation software(Maya)
Opus
Well, got XP loaded and ran a couple benchmarks: PCMark2002 r
Well, got XP loaded and ran a couple benchmarks:
PCMark2002 results:
CPU 5270
MEM 5968
HDD 1000
On this hardware:
ASUS P4T-E, bios 1005, CYP clock chips
P4 1.6a retail HSF @ 133 fsb = 2.13GHz 1.50v
Samsung PC800 256Mx2=512M @ x3 = 399MHz
ATI Radeon 7500 video
pair of IBM 80GB 120GXP Series hard drives
on a Promise FastTrak 100 TX2 mirrored (raid1)
Not too shabby, eh?
SiSoft SANDRA results were all mostly equal to or better than a stock P4 2.0GHz reference except for the hard disk which was just slightly lower than the typical 7200rpm ATA100 reference disk, probably due to waiting on the Promise card to write to both disks.
Looks like I achieved my goal of getting 2.0a GHz performance equivalence out of an OC'd 1.6a and it's rock solid stable.
I haven't yet installed any audio hardware or software but will be messing around with that over the remainder of the weekend.
BTW: using the new machine to post this message :cool:
Good work Opus! In this statement, lies the answer to my unplea
Good work Opus!
In this statement, lies the answer to my unpleasant experience:
"Unlike the 40-pin IDE cable, the 40-pin 80-conductor cable is orientation specific. The connectors of the 80-conductor cable are color-coded: blue for the host connector, black for the primary or master disk drive, and gray for the secondary or slave drive."
Greetings! Great thread here. Take make a long story short, afte
Greetings! Great thread here. Take make a long story short, after struggling with various sound card problems and slow IDE transfer rates, I've decided to ditch my current AMD setup: Gigabyte Ga-7DXR (AMD 761 NB - VIA686b SB), AthlonXP1800, 512mg PC2400 Cosair Ra).
I have a new barebones coming in next week: Abit TH7II-R (Intel 850 -it has raid but I need the extra IDE channels and I don't want to plug in another PCI card. P4a1.6, 2X512 PC800 Rambus).
Anything wrong with this setup? I do intend to try some overclocking so I'll be outfitting the CPU with a better fan/heatsink combo, and at least one good case fan blowing across the memory sticks.
Hey Keith System sounds rocking...only problem being that the 8
Hey Keith
System sounds rocking...only problem being that the 850 chipset has a data corruption bug when using files over 90MB...so be careful!
The 845 chipset is actually a little faster especially due it using DDR memory. The bus speed is faster.
I have been really thinking about the whole RAID thing myself. Might be time to invest in an ATA133 controller card :p
Opus
Originally posted by Opus2000: Hey Keith ...only problem being
Originally posted by Opus2000:
Hey Keith
...only problem being that the 850 chipset has a data corruption bug when using files over 90MB...so be careful!
The 845 chipset is actually a little faster especially due it using DDR memory. The bus speed is faster.
Actually the i850 chipset problem is that the PCI bus max bandwidth can get capped at 90MB/s and slow down under memory-write-to-PCI operations. The old data corruption issue was only with early released batches of the chipset and only happened when you used bandwidth-hungry PCI video cards. About the only real world way to really saturate the PCI bus to make this problem show up nowadays is if you have multiple UltraSCSI-160 adapters with lots of fast disks on them and then run the snot out of them writing to them concurrently. Anybody who would even consider a PCI video card in an i850 machine needs to get drug tested anyway. [[url=http://[/URL]="ftp://download.inte…"]Here's Intel's explanation (see the errata section)[/]="ftp://download.inte…"]Here's Intel's explanation (see the errata section)[/] and here's a snippet of another explanation from an arstechnica forum post:
It's a very specific bug, it's unidirectional, affecting a specific sustained transfer from memory to the ICH's PCI bus. It doesn't affect transfers in the other direction, from PCI to memory. It's only from the MCH to the ICH's PCI bus - it doesn't affect IDE according to the errata (IDE according to the block diagrams doesn't sit on the PCI bus). The bug does not affect the P64H when mated to the 860's MCH (or they failed to document it, but I doubt Intel would overlook that if they already tagged the one for the ICH). In any event, the 845 doesn't have the bug.
From this explanation, it sorta applies to me in that I'm using a PCI bus raid card for my main disk array, but since it only applies the ~90MB/s throttle cap when really doing DMA transfers from memory to something stuck in a PCI slot, it's really a non-issue because I know the grand total bandwidth of all my PCI devices when being written to from memory in DMA transfers, will likely never even reach peaks approaching 90MB/s. The IDE they're talking about is the on-motherboard main primary and secondary IDE controllers, which are not wired thru the PCI bus at all, so thru these IDE interfaces, you get the full ATA/100 specs read and write. If the mobo has on-board raid controller like the ABIT TH7-IIR, that's another story because the raid controller chip *is* wired to the PCI bus just as if it were plugged into a "phantom pci slot".
Still, the i845D chipset is reported to have the full 133MB/s PCI bandwidth unhindered in all operations.... at least that's what is theoretically known to be the case as of this date. Who knows what bugs might be as of yet undiscovered since these chips are just about as complex and sophisticated as the cpus themselves, and Intel has such a grand reputation of making such bugfree chips over the years.
Total overall memory bus thruput in an i845D chipset motherboard is still slightly lower than in an i850 rambus system, but if in a DDR system you use good CAS2-2-2 PC2700 memory and can OC the FSB, you'll be approaching the max thruput territory of a rambus system and you have less "memory latency" (whatever the hell that means) and maybe precise benchmarking can really tell the difference, a human being using the thing sure won't... he'll be too damned impressed by the speed of either an 850 or 845D chipset machine to even care. :D
Aha! Less 'memory latency' with DDR. I knew I had heard that Ram
Aha! Less 'memory latency' with DDR. I knew I had heard that Rambus had some undesirable aspect. Of course, this may be irrelevant in the scheme of things, but as I now recall, the subject came up within a thread on latency at cubase.net... Basically, they were saying how to get systems so low (1-3ms latency) that you could indeed monitor 'live' thru fx and play VST instruments in real time. (Using RME or Nuendo converters at least). In that situation, some people seemed to think that Rambus memory latency might come into play. (Of course, if you don't aspire to super low latency -- 50ms or more is cool because you monitor only thru the coverters' real time software -- this probably isn't a relevant issue for you.)
Now about the i850 bug, if I'm understanding this correctly, it sounds like it could be trouble if you had Audio disks on a Raid card (ATA/100 or 133) and you decided to transfer a large file from your IDE boot drive to your Raid drive(s), provided you had a nice fast IDE. Say that new WD 120gigger with 8mb cache, 7200rpm. So you begin writing and anything that happens to get chunked above 90MB in the first sec is lost? But it would probably be okay to run this setup as long as your Boot drive was, say, tuned down to mode2 (ATA/66), so it would never push more than 90MB per second. Of course, though, I suppose whatever vcache you had could be sitting there waiting to burst as well, potentially also causing trouble. I don't know... I just don't like the idea of having to *reduce* PCI throughput to avoid trouble, so the 845D is looking better and better.
So coming back to the 850 versus 845 debate (aka, DDR vs. Rambus memory), then, it seems both the Rambus latency issue combined with the 850 PCI bug issue puts the 850 at a certain disadvantage, to be weighed against the 850/Rambus's advantage of higher memory bandwidth throughput.
So the next question becomes: How significant to us in audio work is the memory throughput issue? Or more specifically, under what conditions could we potentially max out the 845/DDR memory bandwidth limit? (I think it's 1.5 ghz/sec or something.) Anyone more knowledgable here care to comment? Could this come into play running, say, a 40 track mix loaded with plugs and nearing max cpu utilization on a 2.2 Northwood? (But nevertheless, if this is indeed a bottleneck, I suppose the 845 could alleviate it somewhat by running really fast DDR memory and hiking up bus/memory speeds a bit?)
I'm also curious as to the effect of NickDriver's technique: Raising FSB to 133mhz, while at the same time lowering PCI and Memory multipliers so they run at correct speed. Seems like that oughta punch PCI throughput really good. I suppose this technique could be done on either 850 or 845 platform... just depends on having a good, reliable mobo for stability.
theDuDe
*too much is never enuff!* :mad:
Not really an issue- there *is not* a (regular, heads/platters)
Not really an issue- there *is not* a (regular, heads/platters) single drive that will even *approach* 90 mb/sec transfer speed, SCSI or IDE, so that bug should never come into play without pretty extensive IDE raid applications. You can go ahead and run everything at ATA100, ATA133, whatever, and the actual performance won't even approach those numbers.
Originally posted by theDuDe:
Now about the i850 bug, if I'm understanding this correctly, it sounds like it could be trouble if you had Audio disks on a Raid card (ATA/100 or 133) and you decided to transfer a large file from your IDE boot drive to your Raid drive(s), provided you had a nice fast IDE. Say that new WD 120gigger with 8mb cache, 7200rpm. So you begin writing and anything that happens to get chunked above 90MB in the first sec is lost? But it would probably be okay to run this setup as long as your Boot drive was, say, tuned down to mode2 (ATA/66), so it would never push more than 90MB per second. Of course, though, I suppose whatever vcache you had could be sitting there waiting to burst as well, potentially also causing trouble. I don't know... I just don't like the idea of having to *reduce* PCI throughput to avoid trouble, so the 845D is looking better and better.:mad:
From reading this the 845 is a lower cost/power efficient chipse
From reading this the 845 is a lower cost/power efficient chipset comapred to the 850 chipset..
http://www.intel.com/home/tech/components/chipset.htm
Whether you will see better performance on the 845 vs 850 is hard to say..I'm not 100% sure that it will reflect on audio applications as much as games and graphic intensive apps. Each app uses memory and the system in a different way..most of the builds for their chipsets reflect on standard multi-media systems and not specific audio systems..indeed Microsoft does indeed have Pro Tools systems and so forth but if they have systems for replicating our PC situations I have no idea....
Intel® 845 Chipset Features & Benefits
FEATURES
BENEFITS
400 MHz system bus- Increased system bandwidth for greater system responsiveness and seamless user interaction
256 bit Internal data paths -4x the internal data transfer capability over previous generations
Intel® Hub Architecture -Dedicated data paths to deliver maximum bandwidth for I/O intensive applications
DDR 200/266 or PC133 SDRAM -Choice of memory technology to support a full range of price/performance requirements
AGP4X interface -High-bandwidth interface for high-quality 2D, 3D, and video streams
LAN connect interface -Multiple networking options taking advantage of Intel® SingleDriver™ technology
Alert on LAN* 1.0 -Emits an alert in case of software failures or system intrusion, even when the O/S is not present or the system is suspended
Dual USB controllers -Two controllers for four ports, delivering a total USB bandwidth of 24 Mbps
U
ltra ATA/100 -Takes advantage of the latest industry innovations in HDD features and performance
Intel® Application Accelerator -Software that helps accelerate boot time and application launch times
AC97 Controller- Excellent audio quality, with up to six channels for full surround sound capability including a simultaneous modem connection
Communications Network Riser Card -Allows flexibility for multiple configurations on a single card to extend USB, LAN, and audio
Low-power sleep mode -Energy savings
PRODUCT
PACKAGE
Intel® Pentium® 4 Processor 478 Flip Chip Pin Grid Array (FCPGA)
82845 MCH 593 Flip Chip Ball Grid Array (FCBGA)
82801BA ICH2 360 Enhanced Ball Grid Array (EBGA)
FEATURES
BENEFITS
400 MHz system bus- Supports 400 MHz system bus for single processor configurations
Intel® Hub Architecture- Increased I/O bus bandwidth allows better concurrency for media-rich applications and multitasking
Dual RDRAM capability -Provides 3.2 GB/s memory bandwidth, balancing performance for the Pentium 4 processor based platform
AGP4X interface -2X increase in graphics bandwidth allows highest graphics performance.
LAN connect interface -Multiple networking options taking advantage of Intel® SingleDriver™ technology
Alert on LAN 1.0 -Emits an alert in case of software failures or system intrusion, even when the O/S is not present or the system is suspended.
Dual USB controllers -Two controllers for four ports, doubling the bandwidth at 24 Mbps
Ultra ATA/100- Takes advantage of the latest industry innovations in HDD features and performance.
AC'97 Controller -Better audio quality, with up to six channels for full surround sound capability including a simultaneous modem connection
Communications Network Riser Card -Allows flexibility for multiple configurations on a single card to extend USB, LAN, and audio.
Low power sleep mode -Energy savings
These are the chipset comparisons on benefits..as you see both are very close in benefits and performance...the main difference is the memory controller...
The 82850 Memory Controller Hub (MCH) delivers dual RDRAM memory channels and a 400 MHz system bus, providing the latest graphics support through 1.5V AGP4X technology. Together these features deliver the highest total bandwidth capabilities to the PC platform.
The enhanced 82801BA I/O Controller Hub (ICH2) delivers twice the I/O bandwidth over traditional bridge architecture and provides dedicated data paths to fully optimize the additional bandwidth. The ICH2 makes a direct connection from the graphics and memory for faster access to peripherals and provides the features and bandwidth required for the extended PC.
And the 845 chipset:
The 82845 Memory Controller Hub (MCH) supports a 400 MHz system bus, PC133/DDR200/DDR266 SDRAM memory, and the latest graphics devices through the 1.5V AGP4X interface.
The 82801BA I/O Controller Hub (ICH2) makes a direct connection to the graphics and memory for faster access to peripherals. It provides the features and bandwidth required for extended PC usage models. The ICH2 is already one of the highest volume PC platform products in the world, supporting motherboards based on the Intel® 815, Intel® 810, and Intel 850 chipsets.
Enjoy
Opus
Okay, then, so the 850 'bug' is pretty much irrelevant in most c
Okay, then, so the 850 'bug' is pretty much irrelevant in most circumstances. (And it'll probably get a software fix eventually anyway.)
After scanning the intel stuff above, looks like they are about the same except with regard to memory type/throughput. So it looks to me like the 845/850 decision all comes down to the issue of, on the one hand, Rambus's higher bandwidth and on the other hand, it's potential 'memory latency'. If you ever plan on running a 1.5ms or so monitoring system (with fx, etc.), I'm thinking the 845 looks better. But I'm still wondering if we'll ever even approach the memory bandwidth under any circumstance, other than benchmarks.
Back to reality, though... I'm looking at the Gigabyte GA-8IRXP cause it's on the RME recommended page, plus has a built in Promise Raid 133. The Asus P4T-E as well as P4B266 doesn't have the Raids... which are nice for loads of drives. (I'd just run them in non Raid, for a few extra IDE drives.)
Ohhhh... wait a second, though... If I go with a Gigabyte mobo, I won't be a party to your exclusive society membership... ANUS was it? Have to form GNUS. But I must say I'd prefer being a GNUS to an ANUS!
theDuDe
This may shred a little more light for anyone else trying to mak
This may shred a little more light for anyone else trying to make the best informed decision on a new P4 machine:
http://
Yeah, we agonize over all this stuff, waver, and finally make a decision. Then a few years later it's all so out of date... makes me wonder why I wasted so much time that would have been better spent on making music! Nevertheless, I'll probably agonize another week or so... ;)
One more thing to throw into the mix: The Sis645 chipset is supposed to have a proprietary 533MB/sec transfer between North and Southbridge, while all 850, 845 (and VIA for that matter) stick with a max of 266MB/sec. Not that those will ever be *achieved*, but one might stand to reason that say, copying files from one drive to another in the background and burning a CD, while simultaneously playing a multitrack mix... well you may stand a better chance of that on a Sis645 board, unless there's some massive negative issues waiting there for us.
A little more new info: P4 2.4ghz is due out next week, OEM c
A little more new info:
P4 2.4ghz is due out next week, OEM cost $560 (should push down the 2.2 and 2.0 Northwoods a bit). Also, the date for release of the i850E chipset (133FSB/533mhz) is set for May 6. Not that they are always on time, but it's better than projecting '3rd quarter' release.
http://www.theregus.com/content/3/24482.html
Think I might just sit tight for another few weeks, then... see how this plays out.
BTW, out of curiosity: My current machine is a Dell 800r, with Rambus and an 820 chipset, which was discontinued... why, I have no idea... the thing has worked great for me just over 2 full years now. Anyway, it runs on a 133mhz FSB. Now why in the world did Intel go 'backwards' in subsequent chipsets... back to a 100mhz bus as now in 845 and 850?
Seems to me that for us running large multitrack mixes, the fullest possible communication between north-south bridges is just as crucial an aspect as floating point cpu power. So at this point, I'm kinda split between waiting for the 850E chipset and just going for a Sis645, which runs a far more powerful North-South communication (16bit @ 266mhz = 533MB/sec, as oppossed to i845 and i850 which is still on 266MB/sec). Plus it supports DDR333, which goes a long way toward reaching Rambus memory throughput, but with less latency of SDRAM.
http://www.sis.com/products/chipsets/oa/pentium4/645dxfea.htm
Link removed
Hell, for that matter the Sis745 (for AMD platforms) does away with North-South bridges altogether, in one big hunk. Transfer from memory bus/cpu bus to PCI/IDE is a whopping 1.2GB/sec! And the boards are cheap... Haven't found any reviews on the Sis745, but its predecessor Sis735 got excellent performance *and* stability marks at Toms Hardware.
Link removed
Shit... just when I think I'm narrowing things down to a decision, it all goes to hell and I'm back where I was!! :mad:
theDuDe
Dude...just to let you know I allready tried the Sis645 chipset.
Dude...just to let you know I allready tried the Sis645 chipset...no go! Didnt bode over well man. Wouldnt run audio for shit! Something on that chipset is very bad indeed! i got that chipset in hopes that the data path speed would be huge but it turned out to be shit! So, dont waste your time on the Sis chipsets just yet my friend!
As far as the 820 recall..Intell fucked up the MTH on that one..that's why! Nice and simple aint it! lol! It works with Rambus ram but try putting SDRAM in it(which it's supposed to accept) and all hell breaks loose!
Anyhue..If the 850E chipst is coming out does that incorporate the 90MB PCI data transfer "glitch"?
Also I had a feeling that prices for the 2Ghz Processors would be going down soon..so I will be buying one in the near future and OC it to 3Ghz! HA HA HA HA HA HA!!!! lol
Yeah, the decision between the 845 and 850 is hard but I lean towards anything that "doesnt" have a bug! Plus I liked the price comparison of the 845...
That's my shtick
Opus
Thanks 'Opus2000' for the forwarning of Sis info. That's just th
Thanks 'Opus2000' for the forwarning of Sis info. That's just the thing... something comes out that looks GREAT on paper, but you don't really know until someone actually takes the plunge, sets it up and hits 'record'!
Nothing I read said anything about the 850 bug. Let's hope it's fixed by then.
theDude
Following is a quote from part of the Intel "errata" .pdf file -
Following is a quote from part of the Intel "errata" .pdf file -
5. Sustained PCI Bandwidth
Problem: During a memory read multiple operation, a PCI master will read more than one complete cache
line from memory. In this situation, the MCH pre-fetches information from memory in order to
provide optimal performance. However, the MCH cannot provide information to the PCI master
fast enough. Therefore, the Intel® 82801BA terminates the read cycle early to free up the PCI bus
for other PCI masters to claim.
Implication: The early termination limits the maximum bandwidth to ~90 MB/s.
Workaround: None
Status: There are no plans to fix this erratum.
6. RDRAM* Device Interface Initiate Initialization Operation (IIO) Bit
Problem: The Initiate Initialization Operation (IIO) bit in the RDRAM device initialization control
management (RICM) register may be cleared by the MCH too early.
Implication: If the MCH clears the bit too early and BIOS immediately issues a new initialization op code
(IOP), the MCH may incorrectly control the RDRAM device CMD signal causing an invalid cycle
to write to the memory subsystem, and the IIO bit may not be cleared. This results in a system hang
during the memory initialization process.
Workaround: BIOS must allow at least 1 µs delay between executing initialization opcodes (IOP).
Status: There are no plans to fix this erratum.
Note that in each case the last line reads "There are no plans to fix this erratum." The similar Intel doc for the i845D chipset has NO ISSUES which are not fixed with latest BIOS update. This was enough to steer me away from i850 chipset for an audio machine - if I ever DO manage to bang on that limit, I want it GONE...
Also, check out the Soyo Fire Dragon Mobo if you want everything on one Mobo including stable over-clocking and lots of control over voltages, speeds, etc - That's the board I'm gonna get if nothing better comes along in the next couple of weeks... Steve
Originally posted by knightfly: Also, check out the Soyo Fire D
Originally posted by knightfly:
Also, check out the Soyo Fire Dragon Mobo if you want everything on one Mobo including stable over-clocking and lots of control over voltages, speeds, etc - That's the board I'm gonna get if nothing better comes along in the next couple of weeks... Steve
When I was still trying to make the decision about i850 vs i845D, I looked at the Fire Dragon too, and ran across several web pages that were being pretty harsh on SOYO for reliabilty problems lately and high failure rate of the Fire Dragon. I wish I'd bookmarked the URLs and since I've taken apart my old system, I can't even get to my old browser history files to see where I'd been to retrieve them. If you're bent on getting a mobo loaded full of excessive amounts of built-in interface hardware, the Gigabyte 8IRXP is also an impressive board with almost the same set of on-board features and seems to be in very good favor with the overclocking gamer crowd. Might be one to give a serious look at. Over at the overclockers.com database, there's quite a few 8IRXP's with Northwoods listed in the upper ranks. It also uses a Promise raid controller chip (which I prefer) instead of the HighPoint that the Fire Dragon has, and hey just like theDuDe says, you can brag about being a GNUS (genius?) instead of an ANUS :p .
I've owned a couple of SOYO mobos (BX chipsets, P3 cpus) in the past and had excellent luck with them... they're still going strong to this day, but never owned a Gigabyte although they seem to have a darned good reputation... but if I had it to do all over again, I think I'd have to go for the ASUS P4B266 and do my own raid with a Promise card. There's just something about ASUS mobos that they seem to be a bit more premium quality above the others and 6 PCI slots is a lot to plug various I/O cards into.... those cards that can then be migrated into other machines later when the mobo, cpu and RAM become forced premature obsolete so quickly.
Hey Nick - Thanks for the input - I wasn't sure about the Fire D
Hey Nick - Thanks for the input - I wasn't sure about the Fire Dragon other than feature-wise, til I read a couple of really positive reviews. I too don't remember where, but one of the reviews described their burn-in tests, and any Mobo that could survive that couldn't be all bad - the Soyo never even hiccuped. Also, over-clocking is reported to be extremely versatile and solid. The board has Firewire and a 10/100 "Nick" (so why wouldn't YOU, of all people, like it? :=) both of which I will use with the new DAW in at least one of its "personalities" - My reasoning is this: this machine is not going to be cheap, so possibly having to get a different Mobo if the Soyo for some reason doesn't work out, or even just buy an outboard card to replace marginal on-board functions, is moot to me. The machine will have 8 80GB drives, 2 GB ram, 3.04 gHz P4-2.2, 450 watt PC Power Cooling supply, Firewire DVD-R and CD-RW, twin 21" monitors, dual boot Win2k/XP, and still to be determined but expensive audio hardware (24/96 16 channels or more) so an extra $150 for a second Mobo (worst case) is worth taking the chance that the Fire Dragon will work 100%, leaving me with nearly everything I need in a DAW except SCSI, and still having 6 PCI slots for expansion. I intend to try very hard to get XP to behave on this machine, and will attempt to set it up with ACPI so I can make use of the extra 8 IRQ's XP supports under this mode - If it all works, everything in the machine can have its own IRQ. What a bizarre concept! I ordered and recieved a manual for the Fire Dragon after trying repeatedly to download it from Taiwan, and the board allows complete MANUAL IRQ setting, plus a bunch more custom settings that should allow ANYONE to get in deep enough trouble never to be seen again... :=)
Bottom line: I'm kinda like the Farside cartoon that showed the two spiders setting up their web across the bottom of the slide in a playground - One looks at the other one and says, "Man, if we pull this off, we'll eat like kings..."
Nice 'puter, knightfly. For full tilt, go for a pair of Apogee A
Nice 'puter, knightfly. For full tilt, go for a pair of Apogee AD8000s!
Originally posted by knightfly:
Implication: The early termination limits the maximum bandwidth to ~90 MB/s.
Workaround: None
Status: There are no plans to fix this erratum.[/QB]
I just read the literature on 82801BA I/O Controller Hub 2 (ICH2) (and if I'm not mistaken this architecture is in both the 845 and 850 setups, right?). Anyway, it actually lists the read and write throughputs as:
Read transfers up to 100MB/sec
Write transfers up to 89MB/sec
So it appears to me that this 850 'bug' as far as limiting writes to the PCI bus to around 90MB/sec, well, in fact the i845 then also has the SAME design flaw in practical terms cuz it's paired with the ICH2 which limits reads to 89MB/sec.
Looks to me like they both suffer the same fate.
check it out here:
ftp://download.intel.com/design/chipsets/datashts/29068702.pdf
(page 3)
theDuDe
So it appears to me that this 850 'bug' as far as limiting w
So it appears to me that this 850 'bug' as far as limiting writes to the PCI bus to around 90MB/sec, well, in fact the i845 then also has the SAME design flaw in practical terms cuz it's paired with the ICH2 which limits reads to 89MB/sec. theDuDe
So it seems they do share the same MCH but it relates more to the memory type rather than the chipset itself. RDRAM vs DDR at that point so it pretty much nails the case shut on that subject.
possible higher latency with RDRAM and a data corruption bug as well, makes ya want to lean more to the 845 than anything huh?!!!!
Opus
Hey, Nick - I looked at the gigabyte 8IRXP, and the Promise Raid
Hey, Nick - I looked at the gigabyte 8IRXP, and the Promise Raid was what steered me AWAY from it at the time. So far, all the embedded Promise chips I've found do NOT support raid 0+1 like their add-in boards do. Now, after all the discussion here and elsewhere, I'm doubting whether I ever would want to use 0+1, just because of the performance hit. So, the Giga board is partly back in the running.
I'm still trying to find a way to interface my Tascam DM-24 board with the DAW at high sample rates, so that needs to be solved first, which may dictate several other parts of the puzzle. I really want to try XP's extra 8 IRQ's in manual mode, so for me at least the i845D is cast in stone. Also, as conservatively rated as the P4a's are, a 2.2 OC'd to 3.04 gHz looks pretty good right now. This means the Mobo needs good OC skills, which the Fire Dragon seems to have - There was something about one of the other "contenders" that sorta sucked for OC, but I can't find it now. The one review of the Fire Dragon I read said they beat the shit out of it and it never hiccuped. As long as I don't ever buy a Mobo from the same guys SOS gets his stuff from, it'll probably be fine... :=)
Sorry 'bout that Opus, it was actually meant in commiseration, n
Sorry 'bout that Opus, it was actually meant in commiseration, not to rub salt... I guess that's where AMD got the name Athlon, because they couldn't spell MARATHON... are they half price because they're all ZEROS and no ONES, or because if you aren't careful they BREAK in half, or just because they're letting you pay the other half in the steam that escapes your ears as you try to make them work, or????? Now I'm sounding mean, even to ME... Evil Stevil, the bad boll weevil...
Did someone say Anus? or Hanus? lmao!! Great title man!! love it
Did someone say Anus? or Hanus? lmao!! Great title man!! love it!! lol
Keep the thermal interface on...I've never had to put another paste on to keep it cooler...it's only when you remove the heat sink from the processor after it's been used that you need to add more paste or replace it.
Glad to see another victim..err, I mean DAW user follow my blind ways!! lmao!
Good luck and let the questions roll!!
Opus