Has anyone any run a Duron with KT133A?

[phsstpok]

I would like to see benchmarks of a Duron plus KT133A at 133mhz FSB vs 100mhz FSB, Sandra memory test, Quake III arena, 3D Mark 2000, etc.

Since I presently own a KT7. Tests done with a KT7A would be of the most interest to me but any KT133A mobo test would useful.

 

[Guest]

What speed do you want them compared at? I have a TBird in my KT7A, but I can put the Duron from my other machine in it to run the test, but at what speed so I can give you results like you want?

 

[phsstpok]

Thanks for your offer to help.

Can you run your Duron at 800 mhz? That would work great if you can because 8x100=800 and 6x133.3333=800.

What I am trying to find out is how much a Duron would benefit from using a KT7A vs a KT7. It would work best if you have both, otherwise the KT7A would have simulate the KT7. On a KT7A is it possible to run a 100 mhz FSB with DRAM at 133 mhz like it can be done on a KT7.

Specifically, I wish to learn two things. 1)Just how much does the Duron gain by the increased FSB, a KT7 (or simulated on the KT7A) running 100 mhz FSB and the KT7A running 133 mhz with both configurations DRAM at 133 mhz. 2)How much more overclockable a KT7A is vs a KT7. I'm not interested in top speed of the CPU but how high the FSB and DRAM speed can go? What I mean is if you start at say 667 mhz, (133 FSB and DRAM), how much benefit do you get boosting FSB and DRAM?

Sandra CPU and memory tests as well as Quake III. 3DMark 2000 would be nice also but I know it takes a lot of time. Just the first two would be enough.

I find it very revealing if two Quake III tests are done. One at 1024x768 with all options high and one at 640x480 with all options low. The first tests the system in typical gaming usage and the second tests the CPU and memory bandwidth.

That's a lot but any info at all would help. I have not yet heard of any KT7A testing with a Duron. Your help is most appreciated.

Thank you.

 

[Guest]

Well I could not my Duron at 800mhz, the even numbered multiplier did not work, so I was stuck with 5, 7, 9, 11. I ran the CPU which is a 700 at 933 (7x133) and 927 (9x103), this was the closest comparison I could get, I don't imagine those few MHZ lower hurt anything, since the PCI bus is overclocked 3mhz to balance it. I ran all tests with the exact same setting in the BIOS except for the bus speeds. Here are the results:
Sandra Memory 927mhz: 417-ALU, 443-FPU
Sandra Memory 933mhz: 559-ALU, 585-FPU

Sandra CPU 927mhz: 2655-CPU, 1294-FPU
Sandra CPU 933mhz: 2669-CPU, 1302-FPU

Sandra CPU Multi 927mhz: 3122-Integer, 4330-Floating Point
Sandra CPU Multi 933mhz: 3140-Integer, 4353-Floating Point

Quake3 High Quality:
927mhz - 1024x768: 88.7fps, 1280x1024: 52.8fps
933mhz - 1024x768: 93.3fps, 1280x1024: 53.5fps

One thing that is consistent is all of these tests is the fact that the 133mhz bus only really helps in tests or programs that need memory bandwidth. For Quake 3, this is mainly testing pure CPU speed and Video Card speed, so bus speed does not effect this much since main memory never gets used to run the test, only video memory. Also, I get better results in Quake 3 by overclocking the video card than by upping the FSB. I have noticed I can tweak the CPU a little more with the KT7A due to the fact that the bus speeds are not limited by the VIA chipset any longer. With my TBird CPU I normally run it at 140mhz FSB, any faster creates lockups. With the TBird I noticed that the best setting was with 1050mhz, 7.5x140mhz, compared to 1066mhz, 8.0x133mhz. The faster FSB gave better memory throughput, plus the CPU performance was almost exact being down only 16mhz. Hope this all helps.

My system config:
Duron 700mhz overclocked to 927/933. 256mb RAM Mushkin Rev3, CAS 2 memory setting in BIOS with Turbo and 4-way interleve settings, GeForce GTS 2 32mb DDR at stock settings.

 

[phsstpok]

Cool, Thanks a lot. That's good enough. Your results are pretty much what I expected

Since I only run games, browse the internet, and do some productivity stuff I guess I don't need to upgrade my KT7. I'll concentrate on upgrading my Geforce256 SDR, instead.

One question, please. With the Duron at 103 mhz FSB was the memory the same speed or higher? I'm guessing the same because I score 468 ALU and 552 FPU with a 100 mhz FSB and DRAM speed at 133.

KT7 + Duron 600mhz overclocked to 900. 128mb OCZ "enhanced" pc-133, CAS 2/3/2, FAST memory setting and 4-way interleve settings, GeForce256 SDR 32mb (overclocked at 135/180)

Thanks again.

Phsstpok

p.s. Odd numbered multipliers. Doesn't that mean one of the L1 bridges is not closed?

 

[Guest]

No problem. Yes, the memory was at the 103mhz, I did not enable the FSB + PCI bus to the memory option. I figured seeing the extreme of both ends would be the best to notice differences.

I will have to check the L1 bridges again, but I have not ever been able to get the Duron to run on even multipliers, but my TBird I can pick and choose everything.