*** Official ASUS P4C800/Deluxe (875P) Thread ***

[senior guy]

Originally posted by: computer

Originally posted by: senior guy
Hey guys, I need some memory advice:

I'm giving my current comp to my son and will be building another one using the P4C800 Deluxe. I wanted to check here before buying a pair of 512mb dual-channel memory sticks... Which ones would deliver the best performance in this board bearing in mind that I'll be running it stock ?

I have another related question... Given that my first memory pair will be 2 x 512mb, is it ok to put in a second pair at 2x 256mb? ...is that is doable? and if so, would doing that adversely affect performance?

Thanks in advance.

Can't give any accurate advice without knowing your CPU. PC3700 is cutting it pretty close to get DDR500 from it. Some may O'clock to that speed (which is 250mhz bus) and some may not. The slower the CPU, the faster the memory has to be due to the multiplier. For a 2.4C, you'd want at least DDR500/PC4000 which @default would give you 3ghz. For a 3ghz CPU, PC4000 is a waste of money, plus will have rather bad performance since you'd be UNDERclocking it. You'd have to get 3750mhz from a 3ghz P4 to reach the DDR500 spec.

computer: Sorry I forgot to mention that I have a 2.8c .....so which pair of 512mb sticks would you recommend for my new P4C800 Deluxe? Remember I'm running stock - just want the very best 'default' performance I can get!!!

 

[computer]

computer: Sorry I forgot to mention that I have a 2.8c .....so which pair of 512mb sticks would you recommend for my new P4C800 Deluxe? Remember I'm running stock - just want the very best 'default' performance I can get!!!

This thread is full of memory recommendations, maybe the past 20 pages or so. PC4000 would be about the tops you'd want to go since that (DDR500 @1:1) would be 3.5ghz on a 2.8 CPU. Not all of them will do 3.5ghz, so you may be better off with really good PC3700 like OCZ Gold. If those won't reach DDR500 and you CPU does happen to reach 3.5ghz, you can send them back to OCZ. If you have a Malay 2.8C, or one you know will do 3.5ghz, then go with PC4000 like Geil Ultra Platinum.

 

[NaughtyHead]

Hey guys, I'm running the Asus P4C800 with a 2.8C and a radeon 9800 pro from Sapphire and it rox! I run Farcry, Halo, and UT2004 at 1280x1024 with everthing cranked. Gotta recommend it. BTW, FASTWRITES???!!! SHAME ON YOU!!! Just teasing, but the conventioanl wisdom is that they cause instability. I haven't tried it yet.

 

[DaveR]

Computer, etc...how hot can I let my 2.4c go with stock cooling? it is about 39C idle now (at 275MHZ).

 

[computer]

Originally posted by: DaveR
Computer, etc...how hot can I let my 2.4c go with stock cooling? it is about 39C idle now (at 275MHZ).

Well, the thermal limits for a 2.4C is 74°C I believe, that's 165°F. Personally I get worried when any of my CPU's get over 110°F (F from now on, Celsius sucks!). While running some benchmarks mine got to 120° and I was shocked since it was running about 78° idle! I guess for most people, I'd get concerned at about 120° or so. Your temp is 102°F which is a bit high for idle I think. Make sure you have top notch TIM and a perfectly flat HS base. It's also rather cool around here so my ambient temp's are low.

The fan also plays a vital role. Not for the obvious, but for aerodynamic characteristics. A SMALLER with the same CFM is more desirable than a larger fan with the SAME CFM since the smaller diameter source will have a higher air pressure. I proved that again a few days ago. I had a 6000 rpm 60mm 38 CFM fan on a base that took a 70mm fan. The 70mm YStech fan was bad. I went to short that stupid thermal sensor and when I touched it, it fell off!! I couldn't find another 70mm fan that had a 50 CFM rating. So, I tried a 60mm w/38 CFM rating. I never got the 70mm fan on the HS unit so I don't know how cool that would have run. I was getting 77° idle temps with the 60mm. I put an 80mm 39 CFM fan on it (with a 70-80mm adapter), and the temps went up about 10° and went up much more and faster than the 60mm unit under load. Note that's about the same CFM. I then tried a 50 CFM 80mm fan and that only lowered them a little, but still not to the level of the 60mm fan. Of course if one is trying to cool a large area, the larger fan may be more desirable. But then you're cooling one spot like a CPU's HS unit, you want that air coming in at a higher faster pressure since the cooling effect will be greater. When you have the same CFM emanating from a given source, and the source's diametric area DEcreases, the pressure of course has to go up to keep the same CFM flow. It's sort of like trying to put the same amount of GPH through a garden hose when its diameter decreases; pressure must then increase as does its output pressure. So, if you have a good HS unit, and you can find a smaller fan that will fit correctly on it with the same CFM rating, depending on how your HS unit is designed, the smaller fan may give you cooler temps. My HS unit is made in such a way that it has sort of a semi-hollow core which can really benefit from a more direct pressure flow of air.

 

[computer]

I also saw something interesting a while back regarding Intel's stock HSF units. Someone on this thread asked if they were all the same for P4C CPU's. I thought they were since they all appeared to look the same, and no one chimed in to the contrary. A few days ago when trying to find the CFM of the stock Intel fan (of which I never could find), I came across an article that said indeed the P4 3.06C HT CPU has a larger HS unit and larger fan than that on the 2.4C. So, it may help some of you to try and get that HSF unit for your 2.4C's. Although they may be more expensive that some good aftermarket HSF units. Also note that even the stock Intel unit has a very respectable C°/W rating, even better than most expensive aftermarket HSF units! So, don't waste your $ on a $30-50 HSF unit that has an even worse C°/W rating than what you had originally. The stock Intel unit is I believe .53 C°/W and some expensive HSF units are worse than that. What I got for mine is a .43 C°/W rating (lower is better) and I only paid $11 for it!! $14 at Newegg, but I got two for $11 each. It is a TI-T707TN.

 

[computer]

Ok, I got that version of Sandra installed that I put up the link for, and got similar results as you guys. Strange that the results on this build are totally opposite as the results on the other build!! Figure that one out. On ALL previous Sandra versions, most of the highest numbers were always obtained when "bypass Windows disk cache" was checked. Not anymore. Also, I showed quite a big difference in results when I checked/unchecked the "use overlapped I/O" box so you may want to mess with that some. On my results, the Promise controller is slightly faster.

On ICH5:
Cache used

Drive Index : 75MB/s
Buffered Read : 55 MB/s
Sequential Read : 100 MB/s
Random Read : 59 MB/s
Buffered Write : 85 MB/s
Sequential Write : 43 MB/s
Random Write : 47 MB/s
Average Access Time : 4 ms (estimated)


Cache BYPASSED

Drive Index : 53MB/s
Buffered Read : 56 MB/s
Sequential Read : 68 MB/s
Random Read : 31 MB/s
Buffered Write : 43 MB/s
Sequential Write : 60 MB/s
Random Write : 49 MB/s
Average Access Time : 18 ms (estimated)
----------------------------------------------

On Promise:
Cache used

Drive Index : 77MB/s
Benchmark Breakdown
Buffered Read : 56 MB/s
Sequential Read : 103 MB/s
Random Read : 60 MB/s
Buffered Write : 88 MB/s
Sequential Write : 42 MB/s
Random Write : 48 MB/s
Average Access Time : 2 ms (estimated)


Cache BYPASSED

Drive Index : 53MB/s
Buffered Read : 56 MB/s
Sequential Read : 68 MB/s
Random Read : 31 MB/s
Buffered Write : 53 MB/s
Sequential Write : 59 MB/s
Random Write : 47 MB/s
Average Access Time : 18 ms (estimated)

 

[computer]

Here is what I got on the other Sandra 2004 Pro build for the WD740. Only one number indicates same result.

.............................................Native Controller ........................................ Promise


....................................Cache used/Cache bypassed .................Cache used/Cache bypassed

Index: .................................. 36878/44739.............................................37398/45942
Buffered Read: ............................ 114.....................................................1478/1977
Sequential Read: ..........................48/65 .....................................................49/68
Random Read: ............................ 17/11......................................................16/11
Buffered Write: ............................76/76...................................................1392/1962
Sequential Write:..........................51/63.......................................................53/62
Random Write: ............................21/19.......................................................21/20
Avg. Access Time ms: ...................2/5.......................................................... 3/5

 

[DaveR]

Thanks Clint. Yes, I do have the stock fan. However, my 39C was yesterday when it was 86 degrees out! Room was warm. Not sure my 39 would hurt anything.

P.S. I am about to kill everything on my Promise. I will do 1 lasd hdtach and sandra and then switch to ICH5R.

Film at 11!

Originally posted by: computer
I also saw something interesting a while back regarding Intel's stock HSF units. Someone on this thread asked if they were all the same for P4C CPU's. I thought they were since they all appeared to look the same, and no one chimed in to the contrary. A few days ago when trying to find the CFM of the stock Intel fan (of which I never could find), I came across an article that said indeed the P4 3.06C HT CPU has a larger HS unit and larger fan than that on the 2.4C. So, it may help some of you to try and get that HSF unit for your 2.4C's. Although they may be more expensive that some good aftermarket HSF units. Also note that even the stock Intel unit has a very respectable C°/W rating, even better than most expensive aftermarket HSF units! So, don't waste your $ on a $30-50 HSF unit that has an even worse C°/W rating than what you had originally. The stock Intel unit is I believe .53 C°/W and some expensive HSF units are worse than that. What I got for mine is a .43 C°/W rating (lower is better) and I only paid $11 for it!! $14 at Newegg, but I got two for $11 each. It is a TI-T707TN.

 

[DaveR]

Clint...I also believe that none of your Sandra data is for RAID, correct?
My test may be worthwhile. I use RAID0 on 2 Maxtor SATA drives. I can get my data and switch to Intel and see what happens.

 

[computer]

Originally posted by: DaveR
Clint...I also believe that none of your Sandra data is for RAID, correct?
My test may be worthwhile. I use RAID0 on 2 Maxtor SATA drives. I can get my data and switch to Intel and see what happens.

Correct, I don't use RAID. All tests were single stand alone drives and my PC4800 setup is going to be one WD740 for now.

Guys, again, Sandra is not a very good nor accurate benchmarking tool. You need to try Winbench and PCmark '04 which are world reknown for their benchmarks. AIDA32 is also another great one.

 

[DaveR]

I did get all of those programs but can't we decide on one that we can use?

I actually do not want all of that software on my system and is, in fact, one reason I want to kill the drives and at that point move them to the ICH5R...without adding all of those benchmark programs. I thought hdtach was an OK one...is one of the above better?

Thruth be told I never use Sandra.

Originally posted by: computer

Originally posted by: DaveR
Clint...I also believe that none of your Sandra data is for RAID, correct?
My test may be worthwhile. I use RAID0 on 2 Maxtor SATA drives. I can get my data and switch to Intel and see what happens.

Correct, I don't use RAID. All tests were single stand alone drives and my PC4800 setup is going to be one WD740 for now.

Guys, again, Sandra is not a very good nor accurate benchmarking tool. You need to try Winbench and PCmark '04 which are world reknown for their benchmarks. AIDA32 is also another great one.

 

[computer]

I thought I stated before that PCmark & Winbench to name a couple are among the "standards of benchmarking programs". ATTO and AIDA32 are also great. You can't ever go by one benchmark program alone. You cannot determine everything until all benchmarking programs are used then you take all of them into account. This is what I have done, and I state again the Promise controller is superior. You can take my word for it, or install them yourself and run the programs yourself.

That Beta version of HDtach is obviously flawed since every time I run it I get different results by over a thousand MB/sec. There is thread on it here and most are having the same inaccurate issues with it. It's the total opposite of v2.62, I mean the total opposite results. So, something is "up" with that. Like I said before, you cannot put any credence into a Beta program.

In the last stable version of HDtach, ALL areas of PCmark '04, ALL areas of the Winbench HD suite, AIDA32, Diskbench, ATTO, file copying tests, Sandra OVERALL, the Promise controller won ALL of those and even blew away the ICH5 in most of them. It was a real blowout in the file copying tests to name one. Only one program called "Performance Test" did the ICH5 have slightly better results. Even if it was just PCmark and Winbench alone that were run and the Promise won, that alone would have been enough for me. The fact that's it's ALL of these programs just makes the decision easier. The build of Sandra I was using may have had some better results on the ICH5 by only 2 or so MB/sec, but when the Promise controller has over 1000mb/sec better buffered results, I think that MORE than makes up only a ~2mb/sec difference which no one would ever even notice in the first place. On the new build of Sandra, in the best case results for both controllers, the Promise still got higher rates in my tests.

You also can't go by shear numbers until the numbers are studied. If one controller wins 9 out of 10 tests but by only a few KB or MB or so each, and one controller wins only 1 out of 10 yet wins it by a MILE, the 9 wins are irrelevant in the overall performance picture.

I for one am more than confident that the Promise controller is going to give my setup greater performance and it's a no-brainer decision for me. But, it's your PC of course and it's up to you what you want to use.

 

[gqcowboy]

Sorry if this is a repost - I have tried searching and reading but its just too many pages. I am trying to set up a hard drive to run off the RAID port as a normal IDE. How do I go about doing this? I have done it before, but after a reinstall of wins XP Pro I am at a lost. What should my BIOS look like? What should be the settings? Do I need to install the RAID Driver? I installed the SATA378 driver provided by the motherboard CD (as noted because I was not going to use RAID). Does the hard drive need to be set as a slave? There will only be one hard drive off this port primarily used for storage.

When I connect the hard drive as a slave to the master I get a windows bubble telling me, "Promise SATA Console SCSI Processor Device." What does that mean? Does that mean I need more drivers? I can't really tell you what version of BIOS I'm running, as the DOS screen just before going into BIOS it says its version 1.0.03 (something along that line). Do I need to upgrage to a newer version of BIOS. Sorry for the stupidity - I'm new to this. I appreciate the help. Thanks for your time.

Oh yeah I'm having trouble getting the USB controller under "other devices" in the device manager to go away. It was yellow question mark, now has developed the exclamation point. I know I need driver, but which one? I did the service pack one but this is the USB controller pertaining to "other devices." Heres a link to what my screen shot looks like - Device Manager Screen Shot. I hope this can help identify my problem.

 

[DaveR]

So, I guess what you are saying is that (except for my having a 64k stripe size) I may as well leave it alone on the Promise? If I change stripe to 32k, I could try ICH5R, but wonder if I should leave well enough alone now?

Originally posted by: computer
I thought I stated before that PCmark & Winbench to name a couple are among the "standards of benchmarking programs". ATTO and AIDA32 are also great. You can't ever go by one benchmark program alone. You cannot determine everything until all benchmarking programs are used then you take all of them into account. This is what I have done, and I state again the Promise controller is superior. You can take my word for it, or install them yourself and run the programs yourself.

That Beta version of HDtach is obviously flawed since every time I run it I get different results by over a thousand MB/sec. There is thread on it here and most are having the same inaccurate issues with it. It's the total opposite of v2.62, I mean the total opposite results. So, something is "up" with that. Like I said before, you cannot put any credence into a Beta program.

In the last stable version of HDtach, ALL areas of PCmark '04, ALL areas of the Winbench HD suite, AIDA32, Diskbench, ATTO, file copying tests, Sandra OVERALL, the Promise controller won ALL of those and even blew away the ICH5 in most of them. It was a real blowout in the file copying tests to name one. Only one program called "Performance Test" did the ICH5 have slightly better results. Even if it was just PCmark and Winbench alone that were run and the Promise won, that alone would have been enough for me. The fact that's it's ALL of these programs just makes the decision easier. The build of Sandra I was using may have had some better results on the ICH5 by only 2 or so MB/sec, but when the Promise controller has over 1000mb/sec better buffered results, I think that MORE than makes up only a ~2mb/sec difference which no one would ever even notice in the first place. On the new build of Sandra, in the best case results for both controllers, the Promise still got higher rates in my tests.

You also can't go by shear numbers until the numbers are studied. If one controller wins 9 out of 10 tests but by only a few KB or MB or so each, and one controller wins only 1 out of 10 yet wins it by a MILE, the 9 wins are irrelevant in the overall performance picture.

I for one am more than confident that the Promise controller is going to give my setup greater performance and it's a no-brainer decision for me. But, it's your PC of course and it's up to you what you want to use.

 

[computer]

So, I guess what you are saying is that (except for my having a 64k stripe size) I may as well leave it alone on the Promise? If I change stripe to 32k, I could try ICH5R, but wonder if I should leave well enough alone now?

Finally the forum is back up. :disgust:

I forgot to clarify again, that I did not do any RAID tests.
ONLY single drive tests. I didn't have two of any one
drive. So, it may or may not be different using RAID, and
since I don't use RAID I don't know anything about stripe sizes
and how much they can affect performance. I know they CAN
affect it a lot, but I don't know which stripe size would be
the best and just how much slower things can be if
choosing the wrong stripe size.

This gets more complicated since it would depend on the type of
computing you do. You can run all the benches you want and
test all stripe sizes to find the fastest, but if your RAID
array is used in such a way that requires either an unusually
small, or unusually large stripe size; which ever stripe size
the benches told you was fastest may NOT be the fastest in your
specific type of computing. FIRST you'd have to determine the
perfect stripe size for your type of PC use, THEN run the
benchmarks using that stripe size on both controllers to find
the fastest controller. FAIK, one specific stripe size may be
faster on a certain controller, while a different stripe size
could be faster on the other controller. Geeeze, I'm
glad I'm not running RAID now! The way I am, I'd run myself
crazy over having to do all that. That's the perfectionist in
me again and I drive myself crazy over being that way.

If you're satisfied the way everything is running, then I
say keep it like that. I only wish I could take my own
advice! I'm constantly trying to scientifically determine
the absolute maximum limits of everything in every type
of configuration so I can create/have the perfect setup.
I tell myself: "C'mon Clint, you're not NASA, it doesn't
matter" but I still do it. :roll: :laugh:

 

[gqcowboy]

Fixed it. Man, I was way off! Make sure you have installed the correct drivers. For my case I had to have the Promise SATA378 IDE controller (only one needed). Remember I'm simply using the RAID (promise) port to be a direct connect to another hard drive for storage purposes. BIOS is easy - under advance tab set the operating mode to -> IDE . Then in windows make sure you don't see any yellow points under device manager, but look under diskmanager to see if your new unallocated hard drive is showing up, if so? Let Windows partition the hard drive to the size of your choice.

Man I was going the hard way about things. Just that's what troubleshooting is all about. Thanks to all that helped

 

[DaveR]

Well, Clint, I may try it anyway. BUT, regardless of stripe size, real world data shows my 10 minute spreadsheet opening in 25 SECONDS on this system!

I think stripe size matters most if your way off, like if you need 128k and select 16k. But, 64k to 32k would make little difference.
I am getting a Liteon DVD-Rom (16x) today and want to add it to my setup. I already have a 52x24x52 cdr drive as Primary master, but think I will make the dvd-rom as secondary master.

I wonder if I could use an 80 pin cable or just grab a junkey older one?

 

[computer]

Originally posted by: DaveR
Well, Clint, I may try it anyway. BUT, regardless of stripe size, real world data shows my 10 minute spreadsheet opening in 25 SECONDS on this system!

I think stripe size matters most if your way off, like if you need 128k and select 16k. But, 64k to 32k would make little difference.
I am getting a Liteon DVD-Rom (16x) today and want to add it to my setup. I already have a 52x24x52 cdr drive as Primary master, but think I will make the dvd-rom as secondary master.

I wonder if I could use an 80 pin cable or just grab a junkey older one?

I never use spreadsheets so I don't if that's really good or not, but sounds like it's good. The Lite-one-up LTD163 is about the best DVD drive out there, at least it was several months ago. That's what I have, but it's not installed since I don't need one!

You can use UDMA/ATA66+ 80 pin IDE cables or the old style 40 pin IDE cables for optical drives, it doesn't matter since they are UDMA33 at best. (BTW, FYI, Plextor has a SATA CDRW drive coming out).

 

[DaveR]

I heard about the SATA dcd. I have so many 52x liteon drives. Anyway, the cdr is not black, but the dvd-rom is. Have not hooked it up yet and will probably just use an old 40 pin.

 

[computer]

I meant to say LTD-163. I changed it. Sounds like you're going to be needing to do some painting. If you're painting black, I hope you have better luck that me trying to find semigloss black paint. I had to start a thread on that here. NO ONE locally carries ANY brand of it anymore! NO ONE, NO BRAND. I guess I'll have to get it online somewhere.

 

[DaveR]

Well, the cdrw is the only non-black drive in here. But, i do think I will leave it alone and put the door and side cover on this new system. I HAVE to stop playing as it is my work machine and it FLYS...I need to use it!

Also, where is Xeon? I do not think I have seen him since we got the new forums.

Originally posted by: computer
I meant to say LTD-163. I changed it. Sounds like you're going to be needing to do some painting. If you're painting black, I hope you have better luck that me trying to find semigloss black paint. I had to start a thread on that here. NO ONE locally carries ANY brand of it anymore! NO ONE, NO BRAND. I guess I'll have to get it online somewhere.

 

[DaveR]

Computer. I put the Liteon cd-rw on Secondary master and the new Liteon DVD-ROM on Primary Master. No problem. However, when the system comes up, it makes the DVDROM 2 drives. One the Liteon 166 and the other some SCSI driver. I never saw that before and wonder if I can just delete the SCSI drive? It all works but who need the 2 drive letters for the DVD-ROM?

EDIT: Dumb, it was Alcohol 120% virtual drive!

 

[omanecleoh]

I am planning a first build. The components (most will be purchased from NEWEGG) are the following:
Case ? Lian Li PC-65
Power Supply ? PC Power & Cooling 425 ATX-Deluxe
Mother Board - Asus P4C800-E Deluxe
Processor ? Intel P4 3.4C (Northwood)
Memory ? Corsair XMS 3200PRO LL
Video Card ? ATI AIW 9800PRO
Hard Drive ? Western Digital Raptor 74BG
CD/DVD Burner ? Plextor PX-708A/SW
Sound Card ? Creative Labs Sound Blaster Audigy2 ZS Gamer (SB0350)

The question I have is with the motherboard. I would like to avoid (if possible) a bios flash. Can anyone tell me if they have purchased the P4C800-E Deluxe motherboard with the latest bios v1016 and if so were it was purchased from.

Thanks,
Jim

 

[DaveR]

Jim, I just got mine from Newegg (and a whole lot more) and it came with 1014. I still have it at 1014. Perhaps computer can tell us if 1016 buys us anything.