Re: ALL the "RUMORS" about Asus A8V Dlx and FX-53 and their "problems"

Here's my PEAK stable 24/7 AIR-COOLED numbers:

9 x 293 = 2636.4mHz 1.61Vcore, HTT1465mHz 5 x 293
(I can peak it a bit higher, like 2723mHz with the 9.5x multi)

5:3 DIVIDER, DDR480 @ 3.13V, 2-2-2-5-7-16 -1
(KHX3000 BH5 not compatible with A8V 1:1 (2:1 DDR400) mode, I'd LOVE to try some OCZ 3700EB and go for 290, 1:1)

Temps are actually QUITE nice with Swiftech MCX HS/F, idle @ 34C and Load @ 43C
I hate to disagree with the likes of Macci, et al. but IMHO
This board SMMMOOOKESSSSSSSSSSS

Oh yeah, was anyone still curious about the agp/pci locks? :stick:

beautiful. :thumbsup:

now if AMD would release the 3300+ 939 which would hopefully be around the price of the 3200+...

Quote:

---

Originally posted by Arkangyl
beautiful. :thumbsup:

now if AMD would release the 3300+ 939 which would hopefully be around the price of the 3200+...

---

Yeah, imagine that! 10 x 300 @ 5:3 DDR480 LDT 5x..........
It'd be a KILLER......

Nice!

Now where's that G ?

WOW REALLY NICE WORK!!

Now that looks like a kick-ass combo.

Looks like the busses are locking, and the board is overclocking extremely well.

My only question remains is what about options vs the NF3 250? Doesn't Nvidia have better memory support and/or better HD/RAID options?

Which BIOS?

thx!
It's the BETA 1005.021..... Bigtoe said there will be a "better" new BETA BIOS out soon....

Will be an interesting test....3D benching faster at 300+/5:3 or 250-260 1:1...........Probably the 1:1 action :D
Yeah, I think the NF3-250/939 boards will be AWESOME! Unfortunately ight now they're all vaporware.........

Looking good charlie :)

I had only 2 hours to play with it and didn't have that particular BIOS available. With the 1005.020 the board wouldn't POST if AGP/PCI lock was activated. It works on that BIOS?
The board also maxed at 230FSB for SPi 1M (w/ Vcore, Vddr mods and extra RAM cooling) - and that was w/ both OCZ 3700EB and KHX3500 (BH5)

Looks like you got a good clockin' CPU there too - excellent :D
Lets see some 3D/SuperPi powah?

Macci,
Here's the sPi 1M, but it sucks because I had to run at 5:3 beacause of the A8V/PC2700 issue.......
This is before the Vcore mod @ 1.8V...Di -25C load and Pre- Vddr mod. Macci, when you get one of these, I'm sure it will SCREAM! Except that you can't use KHX3000 :(

C

here's superPi at 291....(DDR476)

here's the MAGIC 300, heheeheh..... she's running nice tonight.
Give me some kind of an idea here, guys....how's the mem b/w? Any chance this'd be faster at 3D than 240, 1:1?

charlie can you run 3D at ~300FSB?

my FX does 7300 at 249 1:1 but that was at over 3 ghz.

That's the highest mem bandwidth I have ever seen. Simply amazing. Nice job Charlie! :rocker:

Quote:

---

Originally posted by macci
charlie can you run 3D at ~300FSB?

---

Dunno'............ let me try.

charlie, I think the multiplier has something to do with the bandwidth of the CPU, I say that simple cause, doing some little tests like running 10x200 gets less bandwidth then 12x200.

of course thats just what the SK8V does for me.
it may be different with the new boards and such.

also, I know OPP has gotten some ridiculous bandwidth shots of around 7600 in sisoft at 255, but honestly I think the key was the high multipliers used for both OPP and Myself when breakking 7300 and higher.

3ghz still aircooled?:confused:

WOW! Great numbers, Kunaak..... really wish I could use my DDR @ 1:1
Macci, heres 3D, 292.............stock board with Vddr mod 3.13v (even NB TIM) not very stable though....2 or 3 game tests max
I think it's the 3.13v holding me back...I need 3.4v or so...

I wish my A8V supported PCI locks :( i know my PC4000 can do over 533MHz.

charlie, what video card is that?
Can you do it with some real card like 9800 etc so that we can see that the AGP lock really works :D
Thanks

Quote:

---

Originally posted by macci
charlie, what video card is that?
Can you do it with some real card like 9800 etc so that we can see that the AGP lock really works :D
Thanks

---

lol.......that's the mighty Xabre 200... don't have any 9XXX VC's here... am waiting for X800's to come in :D
Funny thing though, 3D runs BETTER when I set it to 75/37 from 66/33. This mobo/CPU+your cascade/X800XT = 17K easy.....18K?

of course, higher agp frquency always helps drago's":D

Quote:

---

Originally posted by koensa
3ghz still aircooled?:confused:

---

I think, he mentioned -25 C somewhere for that speed.

where did you get your board+chip Charlie?

get some ram with SPD label of PC3200 or higher instead of the KHX3000 with SPD label of PC2700 might help ?

Quote:

---

Originally posted by charlie
here's superPi at 291....(DDR476)

---

Nice results charlie. Goes to show how things have progressed.

:D

I can't wait to get one of these under my Prommy :).

Quote:

---

Originally posted by macci
Looking good charlie :)

I had only 2 hours to play with it and didn't have that particular BIOS available. With the 1005.020 the board wouldn't POST if AGP/PCI lock was activated. It works on that BIOS?
The board also maxed at 230FSB for SPi 1M (w/ Vcore, Vddr mods and extra RAM cooling) - and that was w/ both OCZ 3700EB and KHX3500 (BH5)

Looks like you got a good clockin' CPU there too - excellent :D
Lets see some 3D/SuperPi powah?

---

If your VIA chipset is a bad VIA chipset then it won't have AGP/PCI lock no matter what BIOS. BIOS 1005.020 should have had an APG/PCI lock, no?

Quote:

---

Originally posted by Speed_Mechanic
If your VIA chipset is a bad VIA chipset then it won't have AGP/PCI lock no matter what BIOS. BIOS 1005.020 should have had an APG/PCI lock, no?

---

Yep, word has it that the early Via Pros where bad, a problem with locks. I guess the latest ones are suppose to work.

OPP

All the early Abit KV8 Pro boards failed when changing to locked AGP/PCI, as did MSI and so it seems Asus. Later Abit boards did not and apparently there are extra resistors behind the VIA chip.

What's this 5:3 ratio Charlie, do you mean 5:4 ? I found through testing with the Abit that to get 5:4 to equal 1:1 I needed an extra 50Mhz on FSB/Hypertransport speed . Testing was with pifast, I did not bother with 3dmark.

This was with a 754 board though with it's lack of memory bandwidth due to being single channel, 939 might have bandwidth to spare so not as badly affected by 5:4 and the high FSB then shines ???

I love AMD chips, I hate AMD chisets :D

Regards

Andy

hey guys....
rig runs nice when cold in the AM....68F in benching room, turned off for 10 hours.....ran this sPi first at 296 and then got 1/2 way through a sPi at 300........
this leads me to believe that there may be merit in sinking mosfets, cooling NB and using more fans on the bench :D
BTW, don't ask me WHY, but I've been back on WinXP for the last day, and the rig runs better AND faster than on Win2K3SVR.

Andy,
It's a quirky thing with the chipset, at 12x it is exactly 5:3, but as you change multi's the divider changes... but I actially LIKE this chipset, to me it ACTUALLY feels like a Canterwood...

Extremely nice board i must say. If the CPU´s weren´t so expensive i´d get a s939 rig right away.

So there is another ratio !

333Mhz to 400Mhz is 5:6 but when I set this I actually get 4:5 , because it takes it off the hypertransport speed and does a dvision on that ie /10, /12 etc etc.

Now you say 5:3 as well. I fear for my Windows Calculator .....

Regards

Andy

nice stuff charlie.

BTW where are you guys wiht the locks getting your boards from? zipzoomfly,newegg, gameve??

thanks

lol.........
Yeah, at 200mHz fsb, it's 160mHz mem speed, at 300, it's what? like 244-245?

OC Kid,
I got mine from zipzoomfly.com.... they take longer to ship than newegg, but it got here ok. But newegg is my fave...if I order by noon, I'll have the order the NEXT DAY!

EDIT: For those laughing at might Xabre 200..... I called Best Buy near here and they actually have stock on X800Pro (the only BB in LA area stocking it)........I may take the drive today :D

C

Ahhh I get it, and 1:1 200 HTT is 200 FSB?

Quote:

---

I called Best Buy near here and they actually have stock on X800Pro

---

:) you know you want to take the drive and pick up some more C02 on the way back :)

sporting some nice mem b/w
and I'm just a "joe six pack" overclocker..... just wait til Macci and OPP get their hands on it! What? 330 benchable? 10 x 330?

The RAM divider or HTT speed doesn't really mean a thing - its all about the RAM speed and timings. 360HTT 280RAM is just as fast as 280FSB 280RAM as long as the CPU speed and timings are equal. This is how it goes on NF150 chipset and I dont see how it would be any different on VIA - the is the part where the RAM controller is anyways..

5:3, 5:6, '166 mode' etc are basically the same thing. All mean that if you boot at 200HTT the RAM will be running at around 166MHz (depending on CPU multiplier).
CPU speed / (CPU multiplier + 2) = RAM speed w/ '5:3, 5:6, 166mode'

Tell me something guys, are the CPU ondie memory controllers still as sensitive to high vdimms? I can't afford to kill any CPUs :p:

thanks for cleaing it up

lookin good:D

Quote:

---

Originally posted by LardArse
Tell me something guys, are the CPU ondie memory controllers still as sensitive to high vdimms? I can't afford to kill any CPUs :p:

---

It's not really an issue with the A8V, because you can't get it over 3.15v anyways... most of the guys "cooking" A64 earlier this year were all at like 3.4v+ on AN50R's

C

Quote:

---

Originally posted by µnrealneo²
Kunaak,

Due to the on-die memory controller, you gain around 150-200MB/s memory bandwidth per 200MHz CPU frequency. :thumbsup:

---

thanks.........so that means someone could break 8000 at about
11.5 x 305

Quote:

---

Tell me something guys, are the CPU ondie memory controllers still as sensitive to high vdimms?

---

I think that the problem was the Shuttle board or sumthing and not just the high Vdimm.
I've killed only one and it happend on AOpen board where my Vbt VR made a 'small' jump - Vbt went from 1.6 to 3.3V => instant CPU death :D
The Gigabyte boards (K8N, K8N Pro, K8NNXP, K8NS Pro) I've used haven't killed a single CPU and I've had Vdimm up to 3.87V. its normally 3.66V - 3.7V for benching.

"small"? lol :D

first of all great stuff...

if i had a money id buy one of these socket 939 FX's

slight off topic question(s)...

1. are the socket 940 and socket 939 FX's similar in performance? i understand that the 940's have to use ECC mem while the 939 does not... how much difference does it really make?

2. both the clawhammer/newcastle(A64) and sledgehammer (A64-FX) work on all the 939 socket boards, correct?

great OCing charlie

whats the issue with the SPD chip on the KHX3000?

and whats the issue with the vDimm circuit on the A8V?

Quote:

---

Originally posted by blinky
great OCing charlie

whats the issue with the SPD chip on the KHX3000?

and whats the issue with the vDimm circuit on the A8V?

---

well here's the deal: KHX3000 sticks have an SPD programming of KIngston Technologies PC2700 256mB
Now the Asus motherboard looks at that and says: Ahhhhh.....slow DDR, so it defaults it to the 5:3 divider (DDR333) There's no way in BIOS to reset it....
The Vddr mod... it works fine up to about 3.15v where it simply locks up in Windows, why? I don't know.

C

hmm, so far, shipping bios, 1003.006, currently at 240Mhz 3dstable, no vddr mod so had ot drop to 5:3 after 220Mhz, 2x512MB of bh-5 cas 2,2,2,5,7,16 . At 240 , assuming no lock and assuming a 1/6 divider, that would be 40Mhz, ain't done any real testing with this x800 pro but what kinda AGP frequency's are they working on?

Also my kv8 pro abit board with stock shipping bios(brought the day they hit uk and musta been almost a month ago, i'd say one of the earliest) did 250Mhz 1:1 2x256mb cas 2,2,2,5 no mods and would also run most dividers at 280Mhz HTT with mem around 200Mhz fine so i assume that lock is working too.

Some info on this board, i got this from various sources, not my bios contact.

Aparantly sales have taken place in europe(UK and Germany) but there shouldn't be any boards available in the US. There is a lock issue but not all boards are effected..so you may get lucky.

Im waiting to see if there will be a new bios out soon..don't mail me for it, as soon as i have it you will i promise.

Charlie. I think you need to start a data base and compile a list of boards that have the lock that works and ones that don't.

We also need hi res pics of boards to find where the 2 resistors are missing that enable the lock.

Macci 1005.020 is a PCI lock bios, your board must be an early one with the resistors missing, once we find where the missing components are and what value you should be able to mod it to lock.

also for some reason you loose 5x ldt with the locks enabled(check sandra). you still get 5x unlocked but even set to x5/1000 in bios with the locks enabled its dropped for 4x so its a bios issue or its been deliberatly added.

Charlie did you try 300ht with a 3x ldt? that may be where your stability went.

Quote:

---

Originally posted by Mongoose420
also for some reason you loose 5x ldt with the locks enabled(check sandra). you still get 5x unlocked but even set to x5/1000 in bios with the locks enabled its dropped for 4x so its a bios issue or its been deliberatly added.

Charlie did you try 300ht with a 3x ldt? that may be where your stability went.

---

Yes...confirmed.... 800mHz HTT
and yes, I hit 330mHz FSB earlier today with ldt at 2x, only for show, though...it takes quite a performance hit.

C

Quote:

---

Originally posted by bigtoe
Some info on this board, i got this from various sources, not my bios contact.

Aparantly sales have taken place in europe(UK and Germany) but there shouldn't be any boards available in the US. There is a lock issue but not all boards are effected..so you may get lucky.

Im waiting to see if there will be a new bios out soon..don't mail me for it, as soon as i have it you will i promise.

Charlie. I think you need to start a data base and compile a list of boards that have the lock that works and ones that don't.

We also need hi res pics of boards to find where the 2 resistors are missing that enable the lock.

Macci 1005.020 is a PCI lock bios, your board must be an early one with the resistors missing, once we find where the missing components are and what value you should be able to mod it to lock.

---

If someone has a board with NO LOCKS, just send me some high rez pic's and I'll compare to find the missing resistors.... just take 4 or 5 pic's covering logical areas on board :D

So say I was running a gig of Corsair XL at 2-2-2 say 220MHz. What memory divider most likely would I be running?

12 x 220 = 2640, 1:1 on AIR works nice :D But it may be interesting to test 10 x 265 = 2650, 5:3..........

C

Quote:

---

Originally posted by macci
I think that the problem was the Shuttle board or sumthing and not just the high Vdimm.
I've killed only one and it happend on AOpen board where my Vbt VR made a 'small' jump - Vbt went from 1.6 to 3.3V => instant CPU death :D
The Gigabyte boards (K8N, K8N Pro, K8NNXP, K8NS Pro) I've used haven't killed a single CPU and I've had Vdimm up to 3.87V. its normally 3.66V - 3.7V for benching.

---

Ok, thats music to my ears indeed! :D

Ah :banana: :banana: :banana: :banana: that means i have a bad VIA board :( Oh well this was just a temp board anyway as i wanted a nForce3 Ultra board :)

could someone take voltage readings of legs 3-4-5 of the RT917B3 chip

Im thinking middle leg 3 should read that of your 3.3v rail and leg 4 should read 1/2 of Vin or leg 1 "vdimm read point"

if leg 3 reads that of leg 1 then maybe the other LM324 is regulating Vin from the 3.3v rail before it hits RT9173b thus your limited to VIN and not of that from your 3.3v rail.

the sk8v used the Lm324+r9173b "but"

leg 3 of r9173b was that of your 3.3v rail

so if leg 3 is not that or "real" voltage set by Lm324 in the corner that might be the issue..

so what Im getting at here is if leg 3 is not that of your 3.3v rail the r9173b chip is setting output based on leg 3 so what I would do is take a direct 3.3v wire from the PSU and connect it to leg 3 of rt917b3 thus your VIN voltage should now be able to go .2v below your VIO/3.3v rail

just a thought

grrrrr.... i was going to buy this mobo and now im afraid to get 1 with no AGP/PCI lock

u guys think ASUS will allow RMAs for boards with no AGP/PCI lock?

Charlie,

Mine is a 45MG0F... and it does NOT support the lock. If i set this option in the 1005.021 BIOS then the Board will not boot.
Tell me what pictures you need and i will send you, so you can compare if there are some resistors missing.

cya

Quote:

---

Originally posted by DazzXP
Ah :banana: :banana: :banana: :banana: that means i have a bad VIA board :( Oh well this was just a temp board anyway as i wanted a nForce3 Ultra board :)

---

same here :( btw ... i can make pics if needed :) , only need to find the driver cd etc of my cam :P

Pictures, well
I really don't know, but the missing resistors MAY be near the NB, so let's just say the area from the agp slot to the socket, to the dimms to the usb ports...
To me it seems as was theorized earlier, all pre-OJ boards do NOT have functioning locks, in addition, ALL UK + EU boards have not worked...'


thx!

here ya go guys the parts to mod 1.01/1.02 boards to support the locks from what Ive heard

and THERE it is!!! thx OC kid!
Do you have a link to the mod, showing values and procedure?

C

http://www.tomshardware.com/cpu/2004...et_939-12.html

they where given this board so it would run faster to compete with the MSI which at that time of review did support the locks.

I think the older boards with the not so good chipsets needed more voltage in areas to hold the locked or asysnc bus freq.

again you would have to try it and see couldnt hurt really

so with that mod i have my locks ??

mayb im gonna try this when i get my solder iron :)

first you have to know what the value is on that SM resistor.... and THG article didn't specifically say that mod was for locks....

C

sketchy