Poll: Ok Post your Sandra scores if you can touch this please :)

[SilverBack]

I'm posting this to get some feedback.
The Anandtech forum's get a lot of attention and has a large reader data base.
Please post your scores regardless if they are better or not.
Here is a link to my scores:

Here is a link to my scores

"The memory bandwidth test that you see here was taken from my current machine. There is currently a misconception that ALL Via chipsets have very low memory scores. Hopefully this will shed a little light on this situation.

The scores were done on the following hardware:
Abit KT7 Socket A motherboard , utilizing a KT133 VIA chipset.
AMD ThunderBird 1 GHz CPU.
Mushkin PC133 Rev 2 memory 128mb

The following settings were made in the BIOS of this motherboard:
Enhance chip performance Enabled
Force 4 way interleaving Enabled
Dram 4k-page mode Enabled
CPU L2 Cache ECC Checking Disabled
Bank Dram Timing Turbo
Dram Bank Interleaving 4-Way Enabled
Delay Dram Read Latch 0.5 ns
Sdram Cycle Length (CAS) 2
K7 CLK_CTL Optimal

Ok that's the settings that this VIA chipset allows. The system is stable and very, very fast. As far as I know of ( and I'm sure I will be corrected ), these scores are the highest memory bandwidth benchmarks by a chipset that hasn't been overclocked to the extreme. For instance a P3 monster running at 1100 MHz using a 160 Mhz memory bus. This was done using a Tbirds 100MHz fsb and PC133 memory support. SilverBack"

 

[Soccerman]

u need to give more options for putting scores.. all I see is an option for YOUR score...

 

[SilverBack]

LOL All I did was follow the forum's guide ,,, Hmmm

 

[Eli]

Yep, I've known that VIA chipsets do not just automatically score worse. It's because of the BIOS settings. Most BIOS' don't allow some of the options to be tweaked..


Very, VERY nice scores BTW!

 

[xtreme2k]

heheh why did u hide the rest of that thing? hheee

 

[SilverBack]

Just saving space on the page it's posted on.... hold on
here's the orignal.


Original pic

 

[Jutny]

i can some close with a 2 year old chipset..... and my p3 1050 beats your athlon 1ghz....

look at this

 

[skemlawn]

Oh ya well my penis is bigger than yours

 

[SilverBack]

Errr Jutny
538/603 is less than 552/629

LOL and like the post said originnaly this is a standard PC133 at 133MHz
And yours is overclocked to 150MHz. You need to look into your system as it should be faster than that and isn't.
Maybe you can tweak your BIOS a little more.

 

[odog]

silverback... imagine the EV6/KT133 and memory @150mhz FSB...

can you say spankage?

 

[rockhard]

Another EV6 bench for u guys.

See sig below.

 

[rockhard]

Silverback

Above bench is done with a T-Bird 1gig @ 9x112fsb+PCI with all the settings the same as urs except the ECC checking option.
I like my system to be rock solid so dont fancy pushin further. Have had it runnin on 120fsb but was a little shaky, probably ram needed tweaking down somewhere?

Was wondering what difference u noticed when u disabled CPU Level 2 Cache ECC checking if any?

Have u tried playing with the CPU Drive Strength option?

Hope u dont mind me picking u for info

rockhard =)

 

[Zeeliv]

Silverback, from the looks of the comparison systems in your full pic you are using an old version of sandra. I'd bet that if you updated they would fall more in line with what most people get (not much of a drop, but there should be some) , and that Jutny might have you beat. It says right on Sandras site and in the prog I believe that comparing scores from two different systems running different versions is pointless.

 

[SilverBack]

Zeeliv,
Actually thats the new Sandra 2000 Version 2000.7.6.49 That is registered.
Others have the wrong one , heh

Hey RockHard great Scores man!!! I'll have to up my FSB a little.
The idea is to let these people know that these TBird systems get great memory benches to!!!
The ECC checking option polls the memory for errors. Your memory stats should be better if disable this option. I'm going to have to lower me multiplier down and try your 9x option. But as stated in the opening of the thread that this was on a non overclocked bus. Sb.

e-phex,
If I could go that high it wouldn't be spankage, it would be annihilation

Thanks for your input guys.
Just imagine in less than a couple of months PC1600 and PC2100 memory will becoming the norm.
Maybe those scores will hit the 1000 limit

 

[Doomguy]

The absolute fastest ram for athlons is Virtual Channel ram. It costs 180 a stick though.

 

[SilverBack]

Rockhard,
Can you please send me an email with a pic of your scores and a little bit about your system.
I would like to post it on the site, provided you supply me with the fastest score.
Thanks in advance
SilverBack
My email addy

 

[rockhard]

thanx silverback

If u can try for 8.5 x 120 u will get around 588/680ish.
Ive had my rig runnin at this and it benched fine but booted me out of quake3
I dont know if its my mushkin rev2 ram or maybe the chipset?
For the sake of 20 odd so marks when my rigs as stable as hell aint worth it IMO.
I was wondering what the CPU to Chipset Drive strength would do?
NE ideas? Maybe this help in higher fsb's?

rockhard =)

 

[SilverBack]

It's set to strong.....

 

[Zeeliv]

Then why does your registered 2000 version have the lame comparison machines and no FPU scores?

 

[utopia]

<< For instance a P3 monster running at 1100 MHz using a 160 Mhz memory bus. This was done using a Tbirds 100MHz fsb and PC133 memory support. >>



Just a few notes;

Why o/c the PIII when you can bench a stock 1.1Ghz PIII? 160Mhz BUS? That would have to be a &quot;6.8&quot; clock multiplier, which does not exist. Closest thing is 7x160 = 1120Mhz.

Current PIII's are NOT able to yield this kind of a score, remember that we have a 133Mhz FSB vs. 200Mhz (DP) here... This benchmark includes all main aspects of a system, CPU-FPU/Cache/FSB/BSB/ETC, and not just memory.

 

[utopia]

Also, if you want to show how fast your VIA platform is run, a real-world benchmark; say Q3 timedemo1, then i will run it on my 1GHz PIII system. This wont prove anything other that my system is faster in Q3, but i am just showing you that your &quot;Memory&quot; score doesnt mean much. High score though!

 

[rockhard]

Silverback

Check ur mail

Couple of 5-600kb bitmaps in there for u.
Sending u CPU one in a couple of mins.
Hope theyre of some use to u,

rockhard =)

 

[SilverBack]

Hey Rockhard,
I took your advice and tried the 9x method..
Scored 578/655
BUT that's a 114MHz bus

Utopia,
As far as the 1100Mhz surely you must know that was made as an exageration. The point was to have a standard benchmark. If you also read the 1st paragraph of the initial post that it's aimed at showing not all VIA chipsets are slow. This one certainly isn't.
Quake 3 demo 1
Settings:
1024 32 bit color, 32 bit textures
High geometry, high textures.
Quake 3 was not altered in any other way. Sound was On.
94.6 fps.

I do have a Prophet 2 GTS 32mb though

 

[SilverBack]

Zeeliv,
That's just the way the new update is .
I don't know why, I didn't write the program.

 

[rockhard]

Silverback

Dunno how the 114 bus?
My KT7 has either 112/115? whacky software?

Hope u got my mails ok,

rockhard =)