sunaiac wrote:

I only own a NOS Asus K7V which I did not get time to test yet (I'm not interested in the KX133, my build is made around the AMD750 because it's voluntarily a full AMD build)
This thread will "force" me to finally test it, I'll report once it's done.
But I don't have much hope, I think only the K7V-T is "supposed" to be thunderbird compatible with models over 700MHz.

Acording to ASUS, the K7V doesn't support Thunderbirds at all (at least officially), the K7V-T does support them, the K7M and K7M-RM does support Thunderbirds though

I have K7M and K7V, but no Thunderbirds. I really just wanted to get the Athlon Pluto/Orion and mess around with the platform and its quirks.

One interesting thing I found with K7M is GeForce FX seems to work better than any earlier GeForce cards on it. More stable. GeForce 256 also has this stutter to it that later cards don't have. I tried GF256, GF2, GF3 and GF4. Maybe the external power connection is the reason for better stability, or they worked out some AGP compatibility issues. TNT/TNT2 work well too. I believe those even run AGP 2x on K7M. The GeForce cards always default to AGP 1x.

I haven't used K7V as much. I got it mainly because of its PC133 support and wanted to see how its AGP compares to K7M.

Really both boards have quite flaky AGP compliance and I recommend Voodoo3-5 if you just want them to work. It's like working with Super 7.

A small correction. EDIT and a large correction of the correcton and even a correction of the correction of the correction. Also lots and lots of edits to try to make my gibberish understandable.

It turned out that I by mistake had installed an Athlon XP 3000+ AXDA3000DKV4E 10.5*200 (I didn't even remember I had this CPU) in the A7M266 instead of the unlocked AXDA3000DKV4D 13*166 I thought I had installed, thats why it was running at 10.5x.

The theory about why some CPUs* with multipliers greater than 12.5x do not post in the Asus A7M266 I wrote about on the last page will have to be revised as I decided to stop caring about what other people thinks! 😁

*Other than the Palomino 2100+, the Thunderbird 1300 and the Thunderbird 1400-100 which have remapped (12.5x = 13x, 5x=14x) multipliers. In fact even the Athlon 1333 and 1400(133) and late Thunderbird steppings in general have the 12.5x = 13x remapped multiplier. It seems the 5x to 14x multiplier can only be used if the CPU is modified to present it self as a genuine 14x100 1400 MHz Thunderbird. Otherwise the CPU will either fail to post when the 5x multiplier is chosen or post at the 13x multiplier even if the 5x to 14x remap should be present in the CPU stepping in question.

The first issue is common and easy to understand. Many old Socket A motherboards can not understand the "high (+8x)" multpler range as they lack the hardware to detect the state of the high/low multiplier range pin on the CPU. If a CPU is wired "high" it should/could still work at the correct speed if it wasn't for the second issue.

The second issue is related to the first issue and it's what causes what is known as the the 13x to 15x "multiplier hole" with Thoroughbred B CPUs that has a default FSB of 133 MHz. The issue is a chipset timing issue but" The Internet" do not seem to understand exactly what causes it.

The exact reason why the Thoroughbred B Athlon XP 2100+ (13x) to the 2400+ (15x) do not work in many older Socket A motherboards is debatable but here is my best guess. The thing making me doubt that I'm 100% correct are reports of some 100 MHz KT133 chipset motherboards also beeing affected while my explanation would only explain the issue with 133 MHz capable motherboards using chipsets like the KT133A and the AMD 760.

___

The Thoroughbred B Athlon XP 2100+ and faster (13x to 15x) with a FSB of 133 Mhz is wired to use the "+8x high" multiplier range. The motherboards without "high multiplier range pin detection" is on a hardware level unaware of the "high range" (even if the BIOS is updated) and thinks the CPU runs at the 5x-7x multiplier. The CPU is requesting a 133 MHz FSB at start up before the BIOS kicks in and the the 133 MHz capable motherboard is hardwired to deliver the 133 Mhz. The 133 MHz FSB is in turn for some reason causing the failure to POST as the same CPU does POST at the same multiplier if hardwired to request a 100 MHz start up FSB.

My thoughts are as follows. As AMD did not plan any CPUs with 133 MHz FSB that used the 5x to 7x multiplier range there were no need to include hardwired 133MHz FSB start up timings for these multipliers in AMDs guidelines for 133 MHz capable motherboards. The motherboard designed to strictly follow AMDs guidelines always use start up timings for a 100 MHz FSB at these multipliers even if the CPU requests and is provided with a 133 MHz FSB.

At the 7.5x multiplier and greater the motherboards are designed to let the chipset adjust their start up timings for a 133MHz FSB CPUs to be able to handle the Thunderbird 1000C and faster and therefore 16x (15.5x is not a valid Athlon XP multiplier) and greater multipliers usually works even with 133 MHz FSB CPUs as 16-8=8x. The 13x to 15x multiplier hole issue could/would/should be fixed by pin wiring the CPU to request a default FSB of 100 MHz at start up.

The update below indicates another probable cause but I will leave this here for now.

___

Update.

The Thoroughbred B 2100+ - 2400+ do not post in the Abit KG7 AMD 760 motherboard either.

Athlon XP-M (X*133) reports 800 MHz at first post in the Abit KG7 but it's to unstable to boot any OS at that speed so I'm not 100% sure what multiplier and FSB it actually uses at start up. The stability is only good enough to get into the BIOS and the BIOS does not show the current speed. If I do not set a valid multiplier and save the BIOS settings at first POST the CPU will not POST again until I reset the BIOS.

I have also found that Athlon XP multipliers are coded with 3x added to the Thunderbird multiplier value. This means that a 13x Athlon XP multiplier is in fact read by an old motherboard as (13 -(-8 + -3)) the same multiplier code as a 2x for a Thunderbird which isn't even a valid setting. This do not really expain why it seems the issue often can be fixed by changing the CPUs FSB ID from 133 MHz to 100 MHz though but it makes my threory above incomplete and opens up other possabilities.

/Update

The above would explain why this can not be solved with a BIOS update as it's happening on a lower level before the BIOS kicks in. It could be solved with a motherboard revision or by not making stupid assumptions based on AMDs guidelines when designing the motherboard in the first place. The latter would explain why not all motherboards are affected. I have not had the time to do enough tests yet but so far all my own findings would be explained by the crack pot theory above.

CPUs with a default FSB of 166 and 200 MHz are not recognized as 133 MHz FSB CPUs and are treated as 100 MHz FSB CPUs, threrefore they work with all default multipliers, even 13x to 15x. The 133 MHz FSB Athlon XP-M could be affected by the issue but at least with the Asus A7M266 the start up multiplier for some reason is detected as 11x and not as 6x as with newer motherboards so I had no issues using these CPUs at all.

I dont know exactly when AMD stopped using hardwired multiplier remaps and intruduced the high/low multipler range selection pin but the Palomino 2100+ seems to use the same 12.5x to 13x remap as the Thunderbird while all Thoroughbred B CPUs uses the new method. It seems the new method was introduced either with the Thoroughbred A or Thoroughbred B, my guess would be that it was intruduced with the Thoroughbred A.

Here is a picture showing the CPU AXDA3000DKV4D running at 13 x 100 but the motherboard thinks it's a CPU with 13x -8 = 5x (or the 5x to 14x remap). The CPU is still at 100 MHz FSB just because this is the first POST with the new CPU and I have changed the FSB to 133 MHz but not yet restarted,

Asus A7M266 BIOS first POST with a Barton 3000+ AXDA3000DKV4D.

Here is a picture of the CPU used for the testing shown above.

Barton 3000+ KQYHA year 2004 week 31.

Here is a interesting CPU I found in my Socket A CPU box, a Barton XP 2500+ with a very very late date code. This CPU would probably be a good overclocker if it wasn't for the fact that it's "super locked" with an 11x multiplier. Perhaps it's worth to try to turn it into an Athlon XP-M later and then cut all L6 bridges to work around AMDs "super lock".

Barton 2500+ AQXEA year 2005 week 28.

Here is another CPU with a very late date code, a Thoroughbred B made in Mars 2004. It seems the damages to the PCB has killed it. It's hard to see the smaller traces on the CPU in the picture but some of them are cut.

Thoroughbred B 2400+ KIXJB year 2004 week 11.

Do someone here is interested in dual Athlon MP systems?

I plan to get a dual Socket 462 board (i found a Tyan Thunder K7XPro S2469 for a good price) someday and a pair of Athlon MPs, also plan to do an Slot A system as well since i don't have any of them

Carlos S. M. wrote:

Do someone here is interested in dual Athlon MP systems?

I plan to get a dual Socket 462 board (i found a Tyan Thunder K7XPro S2469 for a good price) someday and a pair of Athlon MPs, also plan to do an Slot A system as well since i don't have any of them

Yes I will probably build a dual Socket A system sooner or later as I already have the motherboard. 😀

I have finished up my testing (for now) of the Asus A7M266 today so I can get back to benching, or at least back to trying to get the ECS Photon AF1 VIA KT600 motherboard to cooperate with the X1950 Pro.

Here is a picture of my beat up Asus A7M266. The board has a couple of somewhat dodgy caps in the CPU VRM circuit and the rest also seemed in need of some "reconditioning". When I first powered on the motherboard it was about as stable as a house of cards. After the motherboard had been powered on for an hour the screen stopped flimmering and the V-core stopped jumping up and down 0.1V. The board still only found the HDD or CD-drive every third POST and both one POST out of 10 but a BIOS flash solved that issue. I really hate these Mosel flash chips, they seem to degrade with time and can not be flashed with Uniflash.

I managed to install Windows and let the computer run the 7-zip bench for a couple of hours and after that the motherboard seemed totally stable, it still has many quirks though.

Asus A7M266 REV 1.04

Today I got a motherboard from the UK I bought on Ebay a couple of weeks ago. It comes from a smokers house and it's in fact a huge pile of ash with a small portion motherboard hidden beneath. I paid £20 for the motherboard with a Barton XP 2500+ and nice cooler. Even with the shipping cost I find it a decent deal as the motherboard seems to be in good shape... except for the ash. Now I can postpone recapping the other nForce2 boards... Procrastinators of the world unite! ... Tomorrow!

The Abit AN7 v1.0

"Back then" overclockers seemed to prefer the Abit NF7-S v2.0 over the Abit AN7 as they often found them selves in conflict with the Asus uGuru hardware monitoring feature. My self I'm glad to be able to see what's going on in a world of old degraded caps! 😀

And finally! The typical Ebay sellers guide to applying Thermal Compound.

The CPU core is raised above the surrounding area, it will easily come in contact with the cooler so therefore it does not need any thermal paste. The core can in fact get damaged by the paste!

Spread ALL thermal compound in the tube around the core while making sure nothing ends up on the core it self. Should you manage to get any paste on the core by mistake then clean it off immediately!

are there any smp socket a systems that allow for highest speed cpus and fsb? i guess this is better tried with k8,no?

Found 4 256mb sticks in my ram stash. Only one works and at cl3 only. Well, eBay, take my money again !

Here are some performance results with the Asus A7M266 AMD 760 chipset motherboard.

The CPU used is an unmodified Barton XP-M 2800+, the maximum "PowerNow! multiplier" this CPU can use is 16x. XP-M CPUs POSTs at the 11x multiplier in this motherboard.

I did not run an overclocked FSB other than a quick tests at 150 MHz, it seemed 2D stable but 3D-tests showed artifacts. I will test FSB overclocking with my other AMD 760 motherboard to see if this is a chipset limitation, if it's just the A7M266 or if it's the X1950 Pro which can not handle a 75 MHz AGP bus.

Barton XP-M 2800+ (16x133=2133 MHz): Super PI 1M

Barton XP-M 2800+ (16x133=2133 MHz): 7-Zip 32M

The 3D benchmarks are heavily bottlenecked by the AMD 760 chipsets memory speed at 133 MHz. The gain between 1466 MHz and 1866 MHz in 3dmark 2001 is only 1000 points.

3dmark can't even initialize the X1950 Pro with the CPU at the 15x or 16x multiplier. This do not seem to be a stability issue but an AGP timing issue and is probably unrelated to the stability issues I saw with the Catalyst 9.3 driver.

[edit]

Ignore these benchmarks, I found out whats needed to be done to make the AMD 760 chipset run much better in 3D and I switched to the faster (at least with the X1950 Pro) Catalyst 9.3 driver. I ran into exactly the same performance issues with the X1950 Pro running in an Abit KG7 AMD 760 motherboard and also the exact same issue with 3dmark even failing to initialize the card at speeds greater than 1866 MHz. This made me try some other video cards using PCI-E-AGP bridges like the Radeon X850XT, the Geforce 6600GT and the Geforce 7600GS, none of them ran at a decent speed and while the system was perfectly stable in 2D at 2133 MHz it wasn't even fully stable in 3D at 1866 MHz.

At that point I thought perhaps the AMD 760 chipset sucks will all cards using these PCI-E-AGP bridges and switched to a Geforce 6800GT but the instability got even worse, this made my tinker with the advanced settings. I knew Asus recommended changing "CPU skew" from a default 3 to 0 which I tried in the Asus board. With the Asus board I diddn't see any real difference going from the default skrew value to the recomennded "0" though, performance was still good in 2D but the 3D performance kind of sucked.

I tried to change the CPU skew to 0 with the Abit board but found this setting amazingly unstable. If going in one direction makes the issue worse perhaps the other direction makes it better I thought and tried the other extreme, setting 7. Suddenly the stability issues were gone and the 3D performance is what could be expected with both the Geforce 6800GT and the Radeon X1950 Pro and the initialization issue with 3dmark is gone.

There is still the issue with The X1950 Pro not running 3Dmark03 Battle Of Proxycon with the Catalyst 9.3 driver when the CPU is running at a speed over 2 GHz just like with the ECS KT600 motherboard. I have inspected the cards cooler and it's not filled with dust, I replaced the thermal paste. I have also added one fan blowing air on the cards power circuitry and another fan blowing air on the PCI-E AGP bridge but the issue remains the same. This seems to be an issue with the combination of this card and the Catalyst 9.3 driver.

The Catalyst 9.3 driver also still produces a more uneven frame rate in 3Dmark 2001 compared to the 7.10 driver though as the memory becomes an even more obvious bottle neck with better video card performance.

The conclusion is that these old motherboards with AMD 760 really needs a hands on approch with faster CPUs and settings I would have thought only could affect stability also affects 3D performance.

edit2

After some more testing I'm now confident that no Nvidia card with PCI-E - AGP bridge will work with the AMD 760 chipset, or at least not with the Abit KG7. I have retested the 7600 GS and 6600GT and also tested a 7600GT and a 7800GS, none of these card works with any kind of stability regardless of CPU skew setting but briged ATI cards seems to work. Even if ATI cards works and performs well I think that it's probably best to use a non bridged Geforce 6800GT with AMD 760 motherboards.

/edit2

[/edit]

Barton XP-M 2500+ (14x133=1866 MHz) Radeon X1950 Pro 512 AGP Catalyst 7.10: 3dmark 2001.

IGNORE these results, I will rerun the benchmarks with this motherboard later. For now I will post the results with the Abit KG7 and I would think the Asus A7M266 will perform about the same.

Barton XP-M 2500+ (14x133=1866 MHz) Radeon X1950 Pro 512 AGP Catalyst 7.10: 3dmark 03

IGNORE these results, I will rerun the benchmarks with this motherboard later. For now I will post the results with the Abit KG7 and I would think the Asus A7M266 will perform about the same.

Skyscraper wrote:

Today I got a motherboard from the UK I bought on Ebay a couple of weeks ago. It comes from a smokers house and it's in fact a huge pile of ash with a small portion motherboard hidden beneath. I paid £20 for the motherboard with a Barton XP 2500+ and nice cooler. Even with the shipping cost I find it a decent deal as the motherboard seems to be in good shape... except for the ash.

Wow you must really like smokers. All I can see in those pics is fine dust, normal for a computer that has been sitting a poorly ventilated but clean house. My stuff doesn't smell like cigarette smoke (except for some of my clothes), and I am a smoker. To get a board to stink (I've come across some really smelly ones too) they must have sat in a late 90's office building where half the workers smoked or an early 2000's internet cafe... or a bar.

Just give her a good rub with some isopropyl and the smell should go away.

On topic: congrats on the AN7.

Also, if you're looking for a good, fun / stable / fast chipset to mess around with, look for a KT880 motherboard. My favorites are the Asus A7V880 and the Abit KW7. Win98 runs great on them too.

kanecvr wrote:

Wow you must really like smokers. All I can see in those pics is fine dust, normal for a computer that has been sitting a poorly […]
Show full quote

Wow you must really like smokers. All I can see in those pics is fine dust, normal for a computer that has been sitting a poorly ventilated but clean house. My stuff doesn't smell like cigarette smoke (except for some of my clothes), and I am a smoker. To get a board to stink (I've come across some really smelly ones too) they must have sat in a late 90's office building where half the workers smoked or an early 2000's internet cafe... or a bar.

Just give her a good rub with some isopropyl and the smell should go away.

On topic: congrats on the AN7.

Also, if you're looking for a good, fun / stable / fast chipset to mess around with, look for a KT880 motherboard. My favorites are the Asus A7V880 and the Abit KW7. Win98 runs great on them too.

I quit smokng a couple of years ago so I can tell dust from ash! 😀

But yea the motherboard did really stink and the cooler was disgusting. The seller of course still got a perfect feedback, smokers gotta smoke! 😀

I will get a nice KT880 board unless I find one in one of my "unsorted motherboard boxes". 😀

Skyscraper wrote:

I quit smokng a couple of years ago so I can tell dust from ash! :) […]
Show full quote

kanecvr wrote:

Wow you must really like smokers. All I can see in those pics is fine dust, normal for a computer that has been sitting a poorly […]
Show full quote

Wow you must really like smokers. All I can see in those pics is fine dust, normal for a computer that has been sitting a poorly ventilated but clean house. My stuff doesn't smell like cigarette smoke (except for some of my clothes), and I am a smoker. To get a board to stink (I've come across some really smelly ones too) they must have sat in a late 90's office building where half the workers smoked or an early 2000's internet cafe... or a bar.

Just give her a good rub with some isopropyl and the smell should go away.

On topic: congrats on the AN7.

Also, if you're looking for a good, fun / stable / fast chipset to mess around with, look for a KT880 motherboard. My favorites are the Asus A7V880 and the Abit KW7. Win98 runs great on them too.

I quit smokng a couple of years ago so I can tell dust from ash! 😀

But yea the motherboard did really stink and the cooler was disgusting. The seller of course still got a perfect feedback, smokers gotta smoke! 😀

I will get a nice KT880 board unless I find one in one of my "unsorted motherboard boxes". 😀

I can't see how there would be ash on the board unless he/she literally used the thing as an ashtray... Great job on quitting, I've been trying to quit for years but keep finding excuses 🙁

back on topic - I highly recommend the A7V880 (despite my general dislike for Asus mainboards). It's a great motherboard and should be relatively easy to find. The Abit KT880 is better but they are notoriously hard to come by since people who bought Abit usually went for nforce 2 boards...

Performance-wise the KT880 has slightly lower memory performance (1-5%) but overall higher CPU scores. AGP performance is about the same on both chipsets.

It's really quite easy to see from the parts if the rig was used inside a smokers house. It's one of the reasons that smoking is never allowed inside my house no matter what.

It's sticky. But ash is a different thing. it may be particles but I don't think those would seep inside a rig unless the ashtray were right in front of an air intake or so.

kanecvr wrote:

back on topic - I highly recommend the A7V880 (despite my general dislike for Asus mainboards). It's a great motherboard and should be relatively easy to find. The Abit KT880 is better but they are notoriously hard to come by since people who bought Abit usually went for nforce 2 boards...

Performance-wise the KT880 has slightly lower memory performance (1-5%) but overall higher CPU scores. AGP performance is about the same on both chipsets.

It's a shame these early VIA SATA controllers are harder to use compared to the Intel ones though.
A7V880 definitely seems to be an interesting board. I did like KT600 though, despite its shortcomings (which mostly stems from its awkward SATA compatibility).

Tetrium wrote:

It's really quite easy to see from the parts if the rig was used inside a smokers house. It's one of the reasons that smoking is […]
Show full quote

It's really quite easy to see from the parts if the rig was used inside a smokers house. It's one of the reasons that smoking is never allowed inside my house no matter what.

It's sticky. But ash is a different thing. it may be particles but I don't think those would seep inside a rig unless the ashtray were right in front of an air intake or so.

kanecvr wrote:

back on topic - I highly recommend the A7V880 (despite my general dislike for Asus mainboards). It's a great motherboard and should be relatively easy to find. The Abit KT880 is better but they are notoriously hard to come by since people who bought Abit usually went for nforce 2 boards...

Performance-wise the KT880 has slightly lower memory performance (1-5%) but overall higher CPU scores. AGP performance is about the same on both chipsets.

It's a shame these early VIA SATA controllers are harder to use compared to the Intel ones though.
A7V880 definitely seems to be an interesting board. I did like KT600 though, despite its shortcomings (which mostly stems from its awkward SATA compatibility).

I think the SATA shortcomming affects P4 and Athlon 64 chipsets too since the late VIA Socket A chipses shares the same southbridges as some of the P4 chipsets and the early Athlon 64 chipsets

KT600 and KT880 uses mainly the VIA VT8237/R southbridge which also other chipsets uses it as well like the VIA P4X533, PT800/880 series for the Pentium 4 or the VIA K8T800/K8T800 Pro for the Athlon 64

I got today another Socket A board with a CPU attatched

is an MSI KT3 Ultra2 (MS-6380E) with the VIA KT330 and VT8235 southbridge, the board i got doesn't have the RAID nor LAN controller, it has the universal AGP slot, supports Barton/Thorton CPUs officially and USB 2.0

Carlos S. M. wrote:

Tetrium wrote:

It's really quite easy to see from the parts if the rig was used inside a smokers house. It's one of the reasons that smoking is […]
Show full quote

It's really quite easy to see from the parts if the rig was used inside a smokers house. It's one of the reasons that smoking is never allowed inside my house no matter what.

It's sticky. But ash is a different thing. it may be particles but I don't think those would seep inside a rig unless the ashtray were right in front of an air intake or so.

kanecvr wrote:

back on topic - I highly recommend the A7V880 (despite my general dislike for Asus mainboards). It's a great motherboard and should be relatively easy to find. The Abit KT880 is better but they are notoriously hard to come by since people who bought Abit usually went for nforce 2 boards...

Performance-wise the KT880 has slightly lower memory performance (1-5%) but overall higher CPU scores. AGP performance is about the same on both chipsets.

It's a shame these early VIA SATA controllers are harder to use compared to the Intel ones though.
A7V880 definitely seems to be an interesting board. I did like KT600 though, despite its shortcomings (which mostly stems from its awkward SATA compatibility).

I think the SATA shortcomming affects P4 and Athlon 64 chipsets too since the late VIA Socket A chipses shares the same southbridges as some of the P4 chipsets and the early Athlon 64 chipsets

KT600 and KT880 uses mainly the VIA VT8237/R southbridge which also other chipsets uses it as well like the VIA P4X533, PT800/880 series for the Pentium 4 or the VIA K8T800/K8T800 Pro for the Athlon 64

You're right! VIA has obviously also made southbridges for Intel boards and those will be equally affected. I forgot about those and I stand corrected 😀

Here in The Netherlands most boards for AMD during the sA and early A64 era were made by either VIA or NVidia (and as VIA usually made the cheaper OEM stuff it tended to end up more in OEMs) and most of the P4 stuff tended to be made by Intel (with a bit of SiS here and there).

I do remember the K8T800 chipsets being affected, but the ones with the VT8237R Plus southbridge had this bug fixed. I actually tended to concentrate on getting A64 boards with this chipset and tended to avoid the ones which lacked the "Plus" bit. Course things are a bit different now as these aren't exactly flooding the local markets anymore (though they aren't rare either, just not stack after stack for sale for peanuts anymore).

Here comes some benchmark results for my other AMD 760 chipset motherboard, the Abit KG7.

I have to say that for a chipset released in October 2000 the AMD 760 chipset has really good performance, in my testing at least. I read some VIA KT266A chipset reviews from September - October 2001 and the AMD 760 chipset got beaten by about 5 percent on avarage by the one year newer VIA chipset. I will have to test this my self as I own a VIA KT266A chipset motherboard! 😀

The issues I had with 3D performance with the Asus A7M266 were just as bad with the Abit KG7 so I had to do some trial and error to solve them. I did an edit in the Asus A7M266 post above and explanied what I did to fix the low 3D performance with the KG7, I think it would work with the A7M266 aswell. The 3dmark 03 score with the X1950 Pro increased to 13000 points compared to 7000 points before when things didn't work as it should! 😀

Abit KG7, Athlon XP-M 2800+ (16x133 = 2133MHZ), 1x1GB memory, ATI Radeon X1950 Pro 512: 3dmark 2001.

While I did solve the initializion issue with 3dmark and the Radeon X1950 Pro 512 I still could not get the Radeon X1950 Pro to run 3dmark 03 Battle Of Proxycon at 2 GHz or higher so I ran 3dmark 03 at 1866 MHz. This seems to be an issue with the Catalyst 9.3 driver and this particular X1950 Pro. I tried to find another X1950 Pro 512 AGP and I did but that card had bad caps and did not want to work at all. I don't know where my Radeon X1950 256MB AGP card is or I would have tried it but I'm not going to dig through every box it could be in and the performance seemed pretty good even at 1866 MHz so I let it be.

Abit KG7, Athlon XP-M 2500+ (14x133 = 1866 MHz), 1x1GB memory, ATI Radeon X1950 Pro 512: 3dmark 03

I did test my Geforce 6800 "Ultra" aswell, it did not have any issues when running 3dmark 03 with the CPU at 2+ GHz! It gets beaten by the X1950 Pro in 3dmark 03 even though the X1950 Pro was benched with the CPU at a lower speed. In 3dmark 2001 the Geforce card is faster than the Radeon when both cards are benched with the CPU at 2133 MHz.

Abit KG7, Athlon XP-M 2800+ (16x133 = 2133MHZ), 1x1GB memory, Geforce 6800 "Ultra": 3dmark 2001

Abit KG7, Athlon XP-M 2800+ (16x133 = 2133MHZ), 1x1GB memory, Geforce 6800 "Ultra": 3dmark 03

kanecvr wrote:

I can't see how there would be ash on the board unless he/she literally used the thing as an ashtray... Great job on quitting, I […]
Show full quote

Skyscraper wrote:

I quit smokng a couple of years ago so I can tell dust from ash! :) […]
Show full quote

kanecvr wrote:

Wow you must really like smokers. All I can see in those pics is fine dust, normal for a computer that has been sitting a poorly […]
Show full quote

Wow you must really like smokers. All I can see in those pics is fine dust, normal for a computer that has been sitting a poorly ventilated but clean house. My stuff doesn't smell like cigarette smoke (except for some of my clothes), and I am a smoker. To get a board to stink (I've come across some really smelly ones too) they must have sat in a late 90's office building where half the workers smoked or an early 2000's internet cafe... or a bar.

Just give her a good rub with some isopropyl and the smell should go away.

On topic: congrats on the AN7.

Also, if you're looking for a good, fun / stable / fast chipset to mess around with, look for a KT880 motherboard. My favorites are the Asus A7V880 and the Abit KW7. Win98 runs great on them too.

I quit smokng a couple of years ago so I can tell dust from ash! 😀

But yea the motherboard did really stink and the cooler was disgusting. The seller of course still got a perfect feedback, smokers gotta smoke! 😀

I will get a nice KT880 board unless I find one in one of my "unsorted motherboard boxes". 😀

I can't see how there would be ash on the board unless he/she literally used the thing as an ashtray... Great job on quitting, I've been trying to quit for years but keep finding excuses 🙁

back on topic - I highly recommend the A7V880 (despite my general dislike for Asus mainboards). It's a great motherboard and should be relatively easy to find. The Abit KT880 is better but they are notoriously hard to come by since people who bought Abit usually went for nforce 2 boards...

Performance-wise the KT880 has slightly lower memory performance (1-5%) but overall higher CPU scores. AGP performance is about the same on both chipsets.

I found an Abit KW7 recently on ebay, but is 81 €... also found an NF7 for 89 € ahtough i found an ABIT KR7A-RAID for 38 €

I'm currently stuck with 4 MSI Socket A mobos and 1 ASUS mobo

Just wondering, can i contribute in some CPU benchmarks, i would try 3DMark, but i don't have a Radeon x1950 AGP

Carlos S. M. wrote:

I found an Abit KW7 recently on ebay, but is 81 €... also found an NF7 for 89 € ahtough i found an ABIT KR7A-RAID for 38 € […]
Show full quote

kanecvr wrote:

back on topic - I highly recommend the A7V880 (despite my general dislike for Asus mainboards). It's a great motherboard and should be relatively easy to find. The Abit KT880 is better but they are notoriously hard to come by since people who bought Abit usually went for nforce 2 boards...

Performance-wise the KT880 has slightly lower memory performance (1-5%) but overall higher CPU scores. AGP performance is about the same on both chipsets.

I found an Abit KW7 recently on ebay, but is 81 €... also found an NF7 for 89 € ahtough i found an ABIT KR7A-RAID for 38 €

I'm currently stuck with 4 MSI Socket A mobos and 1 ASUS mobo

Just wondering, can i contribute in some CPU benchmarks, i would try 3DMark, but i don't have a Radeon x1950 AGP

I have also seen that ABIT KR7A-RAID for 38 euro on Ebay, it looks like it has some bad caps. "K-systems" (Ebay shop in Bulgaria) has an Asus A7V880 motherboard for about the same price but that board has both bad caps and a beat up DB-25 LPT port.

I kind of stumbled on a KT880 motherboard while looking for something else so I luckily don't have to search for decaying (but still expensive) carcasesses on Ebay.

The X1950 Pro is probably not going to be the only video card used for the CPU scaling benchmarking as I ran in to too much issues. As the X1950 Pro is a video card sunaiac owns it will probably be used to compare Socket A to Slot A unless sunaiac owns some other rather fast AGP card, preferable DX9 but perhaps even a Geforce 4 Ti would be adequate.

The Geforce 6800GT/Ultra seems like a good DX9 card to use for the CPU scaling tests as the AGP version do not use one of those craptastic PCI-E - AGP bridges. It should also run on most motherboards so it could be used for chipset performance comparisons aswell.

I though that I should see if those reviews back in 2001 was correct when they found the VIA KT266A chipset to be about 5% faster than the AMD 760 chipset.

I'm not going to do a serious chipset comparison now, that will come later after I finally have managed to do all the CPU scaling tests. This is more of a quick check to see which of my motherboards actually works as a preparation. Sourcing new motherboards if need be takes some time if one don't want to pay 30 - 50 euro + shipping for each motherboard (often littered with bad caps) from Ebay.

This is my VIA KT266A mother.. ehm okey VIA KM266 motherboard but it transforms into a VIA KT266A motherboard as soon as I stick an AGP card in it. There is also the first version KT266 without A but it was kind of a failed chipset with lots of issues and it soon got replaced by the KT266A revsion. The K7VM2 provides the option to use SDRAM instead of DDR which could be interesting for apples to apples DDR vs SDRAM benching.

ASrock K7VM2 REV 2.02

When I hooked up the motherboard I ran in to trouble, my trusty Seasonic SS-500HT did not like this motherboard. When clicking the "power button" the PSU would only run for a single second before the protection circuits kicked in, a really good sign. I looked for shorts or anything else amiss but found nothing so I switched to the older dumber Chieftec 350W PSU. The Chieftec had an easier time getting along with the ASRock K7VM2 and the system posted at once and when entering the BIOS I found that the motherboard had identified the CPU correctly.

https://www.youtube.com/watch?v=n65vdB28ASY BIOS.

There is also a hardware monitor so I could make sure the motherboard was providing the correct voltage of... 1.85V!?... wait a minute...

Somewhat confused I consulted the provided ASRock instructions video (Well worth a watch).

https://www.youtube.com/watch?v=Eye47G6EcNA

Luckily there is also this guy who shows how to properly set up the ASRock K7VM2...

https://www.youtube.com/watch?v=n65vdB28ASY