PDA

View Full Version : Sandra 2002 memory scores



Greebe
7th June 2002, 13:34
Just upped my FSB to 187mHz, mem 1:1
using Samsung PC2700 DDR.

As above:

RAM Int Buffered aEMMX/aSSE Bandwidth: 2406

RAM Float Buffered aEMMX/aSSE Bandwidth: 2191

:D

Tempest
7th June 2002, 14:43
Nice numbers... Just wait until some Rambus guy comes around :)

I've got an Epox 8K3A+, this board isn't exactly notorious for its huge mem bandwidth but I manage 2036 and 1951 with my 512 meg Samsung PC-2700 stick at CAS2 Turbo settings, standard mem & FSB speeds. Haven't tried overclocking yet cause the speed is currently more than enough for me (and my XP 1800+ is still multiplier locked).

KeiFront
8th June 2002, 02:43
KT7A FSB speed 140
(133Mhz SDRAM Cas 2)

Int ALU/RAM 620 Mb/s
Float FPU/RAM 671 MB/s

Kastuvas
8th June 2002, 09:20
RAM Int MMX 298Mb/s
RAM Float FPU 304 Mb/s

Hehe how slow is that.

Novdid
8th June 2002, 16:26
Give me a link so I can download the piiiiiiiiiiiiiiip!!!:p

Greebe
9th June 2002, 08:59
http://www.sisoftware.demon.co.uk/sandra/index.htm

Indiana
9th June 2002, 09:31
The EPoX 8K3A gives me
Int ALU / RAM Bandwith 2271 MB/s
Float FPU / RAM Bandwith 2130 MB/s

at a quite moderate FSB of 144MHz (I haven't unlocked the AthlonXP yet....)

Greebe
9th June 2002, 09:37
Yeah the KT333 sure does boast some impressive #'s, but it lacks in PCI bus bandwidth and it's.... uhhhh well... uhhh VIA (buggy)

Tempest
9th June 2002, 10:22
It may be buggy...
But I've been running my 8K3A+ board 24/7 with minimum reboots for over a month now and I've not experienced one single crash. I'm running the notorious troublemakin' duo of WinXP + SBLive (+ HighPoint RAID 0 and a Marvel G400-TV with unsupported drivers) so it's stable enough for me :)

bootmeup
9th June 2002, 11:22
RAM int buffered aEMMX/aSSE Bandwidth 1285 MB/s
RAM Float Buffered aEMMX/aSSE bandwidth 1282 MB/s

SDR Sdram with SiSoftware Sandra Ver. 2002.1.8.59

Jake
11th June 2002, 09:29
Originally posted by Greebe
Yeah the KT333 sure does boast some impressive #'s, but it lacks in PCI bus bandwidth and it's.... uhhhh well... uhhh VIA (buggy)

Old via drivers/chipsets, yes.... lately no, I would say.

Cheers,

Jake

Jon P. Inghram
11th June 2002, 11:50
ASUS TUSL2-C, Tuatalin Celeron 1.2@1.38, PC133 SDRAM @ 115 MHz CAS2:

RAM Int Buffered iSSE Bandwidth 778 MB/s
RAM Float Buffered iSSE Bandwidth 763 MB/s

Tom
11th June 2002, 20:40
Originally posted by Tempest
Just wait until some Rambus guy comes around :)
http://members.shaw.ca/tomsplace/misc/SiSoft.jpg

MetalCartman
11th June 2002, 22:34
P4 1.6A (Northwood) @2.4GHz (FSB150)
512MB Infineon DDR-SDRAM CL2 PC2100

RAM Int Buffered iSSE2 Bandwidth 2139MB/s
RAM Float Buffered iSSE2 Bandwidth 2139MB/s

Just did a quick run under XP, without turning off Norton Antivirus and other stuff.

Sandra tells me, that I have ECC-RAM although I'm sure I bought non-ECC memory, because ECC would have been too expensive...any ideas someone?

Regards,
Cartman

Greebe
14th June 2002, 12:29
metal, how many chips on the mem?

MetalCartman
17th June 2002, 10:44
Greebe, I have to open my rig to count them. Will do that as soon™ as my G/F allows that ;)

btw, I like your sig....but who is Dr. Suess?

xortam
17th June 2002, 15:08
... sadly, the question is "Who was Dr. Seuss?"

Greebe
17th June 2002, 19:28
A genius in his own right, shame he past away not long ago :(

Google links to Dr. Seuss (http://www.google.com/search?as_q=&num=10&hl=en&ie=UTF8&oe=UTF8&btnG=Google+Search&as_epq=Dr.+Suess&as_oq=&as_eq=&lr=&as_ft=i&as_filetype=&as_qdr=all&as_occt=any&as_dt=i&as_sitesearch=&safe=images)

Dr Mordrid
22nd June 2002, 12:43
Originally posted by Jake


Old via drivers/chipsets, yes.... lately no, I would say.

Cheers,

Jake

Not really, unless you think the KT-266 and the KT-333 are "old".

I have a device in review here that cannot use all its features on a VIA KT-xxx chipped mainboard, even with an AthlonXP 2000+ or faster, due to PCI bandwidth limitations. This has been a consistant finding throughout the tests, and not just here. This using the latest VIA drivers, including the latest betas.

This is the case with numerous VIA chipped mainboards and not just one. An example is the Gigabyte GA-7VRX I've been testing with, which is by all accounts a very good VIA chipped mainboard and benches very fast in most tools.

So much for benchmarks :rolleyes:

Put the same CPU and other hardware into an el'cheapo ECS K7S5A or ECS K7S6A (SiS 735/745 respectively) and the thing takes off like a rocket.

Now what's your explanation?

Dr. Mordrid

Dr Mordrid
22nd June 2002, 12:55
Originally posted by MetalCartman


btw, I like your sig....but who is Dr. Suess?

You HAVE to be joking?

Unbelievable....even for a European :D

Dr. Mordrid

Jake
22nd June 2002, 22:50
Doc,

I am aware that some of the competitors chipsets have a higher bandwidth. I was commenting on the quality of the drivers. I haven't had a single stability or configuration issue with the KT333 or KT266.

I would really love to see some Sandra numbers from you. Is that possible?

Here are mine:

RAM Int Buffered aEMMX/aSSE Bandwidth: 2032

RAM Float Buffered aEMMX/aSSE Bandwidth: 1930

for the record I'm running my A7V333 memory timings at 2.0-2-2CL 1CMD, Asus setting "Turbo" (135 Mhz fsb). My multiplier isn't unlocked so I haven't played much with the FSB.

Regards,

Jake

Jake
22nd June 2002, 23:00
Here is a screenshot of my Motherboard settings...

/Jake

roadie
25th June 2002, 04:11
lol i have the best score!!!

195
202

this is 192mb cas2 66mhz sdram on my mobile 440bx board.

beat this:)

MetalCartman
25th June 2002, 04:22
You HAVE to be joking?

No, not actually....

But I realised that I've missed something until I checked out the google.com search. It's a pity, that great work from overseas (like that of Dr.Seuss) is not very well known here...

Thanx! :)

cbman
26th June 2002, 23:28
Originally posted by roadie
lol i have the best score!!!

195
202

this is 192mb cas2 66mhz sdram on my mobile 440bx board.

beat this:)

Hah Hah... got you beat big time.

~100
~100

AMD K6-2 300, 512MB PC133@100. ALI M1541 Chipset Mainboard.

Novdid
27th June 2002, 02:59
That's the best score I've seen in a looooong time...:D

mbrock
27th June 2002, 18:45
Originally posted by Dr Mordrid


Not really, unless you think the KT-266 and the KT-333 are "old".

I have a device in review here that cannot use all its features on a VIA KT-xxx chipped mainboard, even with an AthlonXP 2000+ or faster, due to PCI bandwidth limitations. This has been a consistant finding throughout the tests, and not just here. This using the latest VIA drivers, including the latest betas.

This is the case with numerous VIA chipped mainboards and not just one. An example is the Gigabyte GA-7VRX I've been testing with, which is by all accounts a very good VIA chipped mainboard and benches very fast in most tools.

So much for benchmarks :rolleyes:

Put the same CPU and other hardware into an el'cheapo ECS K7S5A or ECS K7S6A (SiS 735/745 respectively) and the thing takes off like a rocket.

Now what's your explanation?

Dr. Mordrid

I have been using VIA chipset for years and I have always found it to be a performance set when combined wit an Abit motherboard. As far as bandwidth goes the Abit KX333 does not start reaching it's potential until around 205 on the FSB. Go to a search at Mad Onion on AMD CPU's and you will see that 90% of the top scores are done using the VIA set.

A sample of VIA kt333 memory bandwidth at only 190X10. My water is not cold enough right now to show what it is at 210X10.

http://68.53.93.81/190x10.jpg

mbrock
27th June 2002, 18:55
200X9.5 kt333
http://68.53.93.81/200x9.5.jpg

Jake
28th June 2002, 01:24
Those are MONSTER numbers mbrock. Can you post some pics of your water rig?

/Jake

mbrock
28th June 2002, 16:28
That is nothing compared to what this chipset is capable of. My friends DDtung, MrIcee and Oppainter from another forum get are getting much better results. The key for my results is the Abit KX333 mother board combined with Samsung PC2700 RAM running at 2.9 to 3.1 volts on the vmem.

Here is a link to some pics. They are posted on another forum where I used to be a member of their 3d team. the system has changed some (mother board, video card, ect..) since these were taken.

pic of system are here (http://www.icronticforums.com/showthread.php?threadid=27226)

Greebe
28th June 2002, 23:15
Now those voltages are just insane... lookin for short lived mem that for sure. Also the limitations of the PCI bus bandwidth with the KT333 chipset is appalling

mbrock
29th June 2002, 11:06
You are correct in the fact that the high Vmem will reduce my ram’s life. I will give you that. Life of a stick of ram is what 10 - 20 years under optimal conditions? As long as the ram is kept cool even if the high voltage reduces its life by 75%-80% it will have well served its purpose. How many people do not upgrade their performance/gaming rig at least every 2 to 5 years? I do not know where you got your info but I would put my Abit KX333 up against your MSI or any other AMD chipset any day. What are you basing your conclusions on? If my PCI, AGP and mem bandwidth is less then I need to change. I have seen nothing yet to prove this as fact.:) Please enlighten me so I can improve my performance.

Marshmallowman
30th June 2002, 23:15
mbrock, via 266a and 333 have excellent memory performance, and good AGP performance....But the PCI performance not very good, most games and benchmarks will not show these limitations.

You will notice the difference between a kt333 and a sis745 system when you are using multiple PCI devices, eg RAID controller and a capture Card...in fact via chipsets limitations show there limits just using a RAID controller.

and crackling soundblaster's arn't fun either

Jake
30th June 2002, 23:56
RAID perfromance looks fine on my A7V333...

/Jake

mbrock
1st July 2002, 04:14
Originally posted by Marshmallowman
mbrock, via 266a and 333 have excellent memory performance, and good AGP performance....But the PCI performance not very good, most games and benchmarks will not show these limitations.

You will notice the difference between a kt333 and a sis745 system when you are using multiple PCI devices, eg RAID controller and a capture Card...in fact via chipsets limitations show there limits just using a RAID controller.

and crackling soundblaster's arn't fun either

Maybe, but my SB live has not cracked and popped sisnce my old KR7 and only because I had my PCI bus so far out of spec with the FSB at 195 the PCI was at 48.75. My KX333 has not had those problems on either the SB or my NIC.

Greebe
1st July 2002, 07:43
You are correct in the fact that the high Vmem will reduce my ram’s life. I will give you that. Life of a stick of ram is what 10 - 20 years under optimal conditions? As long as the ram is kept cool even if the high voltage reduces its life by 75%-80% it will have well served its purpose. How many people do not upgrade their performance/gaming rig at least every 2 to 5 years? I do not know where you got your info but I would put my Abit KX333 up against your MSI or any other AMD chipset any day. What are you basing your conclusions on? If my PCI, AGP and mem bandwidth is less then I need to change. I have seen nothing yet to prove this as fact. Please enlighten me so I can improve my performance.


18+ years OCing computers and experience as an EE in doing so... where do you get yours?

BTW as was already stated the 745 chipset whips any VIA chipset arse. You should realise that not everything based on today's best bechmarks will test this... just like peak fps on a GF4 doesn't mean it doesn't stutter or lacks when the going gets tough.

Dr Mordrid
1st July 2002, 07:52
Where the PCI performance difference really shows is when you have PCI devices that require a high bandwidth like the RT.X100.

On the KT-266/333 systems this chokes, often causing playback stutters and the downright failure of some functions like realtime export to HDD or IEEE-1394 when you have multiple effects applied to the project.

This problem on the VIA chipsets is so bad that Matrox is not recommending VIA KT-266/333 chipsets for use with the RT.X100.

Where the SiS chipsets shine is because they do multi-threaded I/O plus they give each PCI slot it's own DMA, I/O etc. etc. to extend the bandwidth.

It works spectacularly as even an el'cheapo ($65 USD) ECS K7S6A (SiS 745) can run the RT.X100 full tilt & then some while a much more $$ board like the Gigabyte GA-7VRX (VIA KT-333) falls flat on its face.

The SiS 645 chipset for the P4 performs similarly and it too has multithreaded I/O etc. and is recommended by Matrox for the RT.X100.

Dr. Mordrid

mbrock
1st July 2002, 08:55
Dr Mordrid

Thank you that was the answer I was looking for. Not some off the wall info like "BTW as was already stated the 745 chipset whips any VIA chipset arse" that absolutely gave me info on what I was asking. Thanks again Dr.Mordrid.


Greebe
Quote:---------------------------------------------------------------------------
"18+ years OCing computers and experience as an EE in doing so... where do you get yours?

BTW as was already stated the 745 chipset whips any VIA chipset arse. You should realise that not everything based on today's best bechmarks will test this..."
End Quote----------------------------------------------------------------------

No I am not an EE and I have not been OCing for quite 18years but I do I do a pretty good job of OCing myself most of which I learned the hard way through trial and error. My KX333 suits my needs perfectly it is built as a hobby machine Benchmarking and Gaming. It performs those two functuions as good as any. I would hope after 18+ years of OCing you would get more than 123mhz on your OC.

Excuse me if I offended you. I used to always use M cards and used my G400 max until I was dissapointed with Matrox not putting out a true performace card last year. Nvidea and ATI was my only choice. I hope the 512's performance increases as new drivers are released if so I will return to Matrox.

Greebe
1st July 2002, 09:06
You didn't offend me at all, but it's obvious your not reading the other threads discussing this. It's you not me that should pay attention... it's 186 mHz (and stated in my sig at 184) not 123! :rolleyes:

mbrock
1st July 2002, 09:37
Excuse me:

"OS: XP, XP 1800+ @1656"

XP1800 stock = 1533mhz
1656 - 1533 = 123mhz

You are correct I have not read the other threads. I come here to MURC mainly to keep up on how Matrox is doing and what new is on the horizonfor them. I came across this thread and decided to post my Sandra mem score. I was really looking for bench marks on the 512.

mbrock
2nd July 2002, 17:02
PCI bandwidth may suck but you have to love the memory bandwidth on the KX333

http://68.53.93.81/210x9.5.jpg

Greebe
2nd July 2002, 18:22
Regardless mem speed cannot make up for lack of bandwidth.

And oh your making fun because I'm not pushing my cpu beyond a 123mHz OC?

Did you possibly think it's because I haven't bothered with trying for more as I'm busy beta testing Parhelia?

Noooooooooooooo :rolleyes:

Silencer
2nd July 2002, 19:30
I don´t trust Sisoft Sandra 2002.
The memorybench shows a bandwith around 1200MB/sec, but I use a i440BX with a PL-iP3/T and a P-III S 1.13GHz @ 1.28GHz (150MHz FSB).
Sisoft Sandra 2000/2001 show the correct score of 590 to 600MB/sec.

mbrock
2nd July 2002, 19:40
I do not normally use 2002 either but that is what this thread was for so I upgraded my 2001 to 2002 just to compare Oranges with Oranges. If you disable the buffering crap the score in 2002 is about the same as 2001.

Silencer
2nd July 2002, 19:42
thanks for this info.
however, now I use wstream for memory benchmarks.

mbrock
2nd July 2002, 19:50
The only rteason I use Sandra is because it seems every reviewer on the web uses it. The program means nothing to me other than it is a way for me to compare my systems mem bandwidth with other OCed systems. :)

Silencer
2nd July 2002, 20:00
yes.. its a popular programm..
I compare my systems with others based on setiscores..
I think, a seti score shows the power of a system more, than any other benchmarks :)

Greebe
2nd July 2002, 20:25
So mbrock you're claiming that you've tweaked Sandra 2002 so the mem #'s come out faster... am I reading you correct?

mbrock
2nd July 2002, 20:33
Not at all. I have twaked my system yes but not for Sandra. Is there somthing wrong with that? Isn't tweaking part of getting a good OC?

Greebe
2nd July 2002, 20:36
If you disable the buffering crap the score in 2002 is about the same as 2001.

What does this mean then?

mbrock
2nd July 2002, 20:51
It means:

When Sisoft changed from 2001te to 2002 they added something called buffering. By doing this the scores for 2002 increased. Buffering is a default setting under options. By disableing buffering in the options it makes the score resemble a 2001te score. Which is much much less. If you look at the posts in this thread that have pics you will see the word buffering just like mine does.

Silencer said he did not trust the 2002 program I simply gave him a way to use 2002 to compare to a 2001te score.

Greebe
2nd July 2002, 21:02
Silencer said he did not trust the 2002 program I simply gave him a way to use 2002 to compare to a 2001te score.

Nor do I and as you said I only use for comparative purposes

mbrock
3rd July 2002, 06:47
Originally posted by Greebe


Nor do I and as you said I only use for comparative purposes

Then why did you start this thread?:confused: Good luck on your testing of the 512. I am sure the 2D quality is great as M always is I just hope the performance in 3D also turns out to be great.:)

GT98
3rd July 2002, 12:40
Some one was asking for Memory Benchmarks for my Asus A7S333 before, but I was having problems with the KingMax Memory I had in and has been replaced with some Samsung DDR333 Memory and now only populating 2 Slots, thus letting me experment with PC2700 speeds later on. Right now these are my benchmarks without any tweaking going on or the like:

RAM Int Buffered aEMMX/aSSE Bandwidth: 1650

RAM Float Buffered aEMMX/aSSE Bandwidth: 1563

I'll play around with the Ram Settings and OC the FSB later this weekend :)

Greebe
3rd July 2002, 13:33
Sorry to say this, but I must... It wasn't intended to be a pissing contest until you chimed in.

mbrock
3rd July 2002, 14:39
OK MR MODERATOR if you say so.

I post a score that rivals rambus.

You crap on my bandwidth.

I ask a serious question about where you got your info.

You give me a bullshit answer of "18+ years OCing computers and experience as an EE in doing so.."

I may be a MINI MURCER here but I am no dummy when it comes to tweaking a systems hardware and OS for maximum performance.

So blame me if you want.

Ant
4th July 2002, 00:47
Quit it.

GT98
5th July 2002, 12:34
Did some more playing around...computer didnt like being at 145 FSB, but seems stable at 144 :rolleyes: Heres my lastest numbers:

RAM Int Buffered aEMMX/aSSE Bandwidth: 1779

RAM Float Buffered aEMMX/aSSE Bandwidth: 1682

I'm running the ram at 1:1 and my Memory settings are 2:2:3:5..if anyone has any tweaks I can do I would appreat it :)

KvHagedorn
7th July 2002, 19:48
I think it makes a lot of sense for Dr. Mordrid to seek the best in PCI bandwidth, since he uses an RT series card on a regular basis. For those seeking higher memory bandwidth but whose needs on the PCI bus are not so great, the KT333 might be the better choice.. I don't see what the big deal is.. if I was using an RT board or a gigabit ethernet card, you can bet I would have gotten one of the AMD760MPX boards, or an SIS board if I was only going single processor. I got a KT333 because the box I just built is primarily a gaming rig. To each his own and to all be happy. ;)

MetalCartman
7th July 2002, 22:34
Amen ;)

SIS is becoming more and more interesting for a server-upgrade...

Greebe
7th July 2002, 22:57
Actually I think the reason for the largest difference in scores is caused by fast writes being disabled on all SiS 745 chipset MB's atm. Have heard that this very well may be addressed shortly. As for your speculation as to why you think VIA is better choice for gamers and the SIS chipsets being better for High demand PCI use ie Video editing/server use may seem true on the surface, but IMHO VIA doesn't deserve to be supported bacause of their shitty drivers track record over the years.

(I put ATi in this boat too)


BTW The scores we're seeing on the SIS 745 chipset is only a tad slower than the VIA KT333's... nothing like the huge diffference in bus bandwidth. Also ~40% less $$!

ALBPM
20th July 2002, 20:37
Well, where are the Rambus guys?????:D :D

Here's some more DDR Goodness

http://mywebpages.comcast.net/pmahl/166mem.jpg

Paul

KvHagedorn
21st July 2002, 00:36
RAM Int Buffered aEMMX/aSSE Bandwidth: 2391
RAM Float Buffered aEMMX/aSSE Bandwidth: 2211

This is an XP1800+ running at 167x10, memory at 167 cas2-2-2 Corsair XMS3000 on the Epox 8K3A+.

Evildead666
22nd July 2002, 06:47
I'd just like one question answered pleeze:
Why is via so poor in performance.
Is it the IDE driver? PCI bus driver?
I have an Abit KT7A-RAID, and i seem to have less IDE overhead than with my BX chipset????

Evildead666
22nd July 2002, 06:48
sorry for posting this here, but people are slagging the via chipset, and i just want to be sure its the mobo...