Clock per Clock Comparison: 680i vs 975X

[BenchZowner]

As ideas keep flowing in my head ( lol ), I thought I should write a article about clock per clock performance of the 4 main contenders in the C2D era :D

Since most people here are overclockers, or should I say benchmarkers ( shows the relation to benchmark numbers, etc ;) ) I thought I should give this part of the article a small ( non visual quality ) presentation.

Information:
I've tested the Intel 975X Chipset & the nVIDIA nFORCE 680i chipset in various benchmarks, by using the same ( exact ) configuration ( software & hardware-wise ) on both tested motherboards.
I've matched ( where applicable ) the memory timings ( and sub-timings ofcourse ) and clock frequencies of both configs. ( the same parts were used on both motherboards )

=======================
Test System
=======================
Motherboard: Asus Striker Extreme ( 680i ) & Asus P5W DH Deluxe ( 975X )
CPU: Intel Core 2 Duo Extreme X6800 ( @ 3.5GHz / 350MHz x 10 )
RAM: OCZ PC2-6400 S.O.E. Urban Elite 2GB Kit
VGA: Twintech GeForce 8800GTX
HDD: Maxtor MaXLine III 300GB

=======================
Configurations
=======================
FSB: 350MHz ( QDR-1400 )
DRAM/FSB: 1:1
RAM Mode: Linked
RAM Sync: Synchronized
RAM Timings:http://weboperative.com/OCIDB/albums/userpics/10033/memset.jpg
HyperPath 3 - GPUex - LinkBoost: Disabled
BIOS Versions: Striker1002 - P5W1602
OS:Windows XP Service Pack 2
Drivers:Intel Chipset Software Installation Utility v8,1,1,1010 - nVIDIA nFORCE v9.53 - nVIDIA ForceWare v97.92

===============================
The results
===============================

http://weboperative.com/OCIDB/albums/userpics/10033/SuperPi1m.jpg

http://weboperative.com/OCIDB/albums/userpics/10033/SuperPi32m.jpg

http://weboperative.com/OCIDB/albums/userpics/10033/3D_Mark2001SE.jpg

http://weboperative.com/OCIDB/albums/userpics/10033/AquaMark3.jpg

http://weboperative.com/OCIDB/albums/userpics/10033/3D_Mark03.JPG

http://weboperative.com/OCIDB/albums/userpics/10033/3D_Mark05.JPG

http://weboperative.com/OCIDB/albums/userpics/10033/3D_Mark06.jpg

http://weboperative.com/OCIDB/albums/userpics/10033/Hexus-PiFast.jpg

http://weboperative.com/OCIDB/albums/userpics/10033/wPrime.jpg

SuperPi 32M Screenshots
http://weboperative.com/OCIDB/albums/userpics/10033/normal_superpi.jpg

Edit:
The tests were run ( x times ) & averaged:
3D Mark2001SE = 2 times
3D Mark03 = 2 times
3D Mark05 = 1 time
3D Mark06 = 1 time
AquaMark 3 = 3 times
SuperPi 1M = 5 times
SuperPi 32M = 1 time
Hexus Pi-Fast = 2 times
wPrime = 2 times

[krampak]

Maybe I've missed it.. 680i is running at 1T or 2T ?

I did a similar comparison but with more mobos and more programs, you can check it here (http://www.todoreviews.net/modules.php?op=modload&name=Reviews&file=index&req=showcontent&id=340&page=8). Both 680i are running at 2T.

[BenchZowner]

Maybe I've missed it.. 680i is running at 1T or 2T ?

You mean Command Rate ( also known as Command Per Clock ).
It's in the configuration section.
Running 2T Command Rate on both chipsets.

Although, I plan on comparing both chipsets at their best possible settings later on ;)

[hawtrawkr]

Although, I plan on comparing both chipsets at their best possible settings later on ;)

thats what ill be waiting to see. good to see a clock by clock comparison though.

[krampak]

Running 2T Command Rate on both chipsets.

Does 975X have Command Rate options ? I guess not.

[BenchZowner]

Does 975X have Command Rate options ? I guess not.

There isn't a Command Rate option available on most 965P/975X motherboards.
They default to 2T :)

[sdumper]

I hate to mention the obvious but at this point it is widely known that for the 680i to match or beat a 975x based mobo you need to go 1t and trim the timings. Also note that your not going to get much beyond 400/1600 bus speeds on the 975x last and most important to anyone using the 680i boards Sli is a deciding factor and your not going to run the 8800gtx cards in Sli on 975x boards even with the modded drivers.

[Speederlander]

Do these runs with a quad, and the better stability of 975 boards will allow higher overclocks and better results on the multithreaded tests. When you test to their limits.

[SSXeon]

Interesting, nice comparison. :D

[BenchZowner]

I hate to mention the obvious but at this point it is widely known that for the 680i to match or beat a 975x based mobo you need to go 1t and trim the timings.

That's right, but would this make it a fair comparison ? Nope.
Max performance per clock between these chipsets will be 'investigated' in another part of this article :)


Also note that your not going to get much beyond 400/1600 bus speeds on the 975x

Depends on your Vmch and luck.
I got a 475MHz stable P5W DH Deluxe without any mods ( stock, rev. 1.03g, 1.85Vmch ).


last and most important to anyone using the 680i boards Sli is a deciding factor and your not going to run the 8800gtx cards in Sli on 975x boards even with the modded drivers.

Yeah, but when it comes to single-card benchmarking...even a 100 points gain can be negligible ( in benchmarking )

[Kristoferr]

Tests prove what we all know already :)
I975x is faster if it comes CR 2T and Single graphics card.

If you want to go rof SLI and you have DDR that does ddr800 with tight timings and 1T (like micron fatbody) then go for 680i

Its simple like that!
Cheers!

[crazyea]

I got a 475MHz stable P5W DH Deluxe without any mods ( stock, rev. 1.03g, 1.85Vmch ).


You mean the P.O.S. in your sig? LOL

[safan80]

You mean the P.O.S. in your sig? LOL

some people expect too much.

[sdumper]

Depends on your Vmch and luck.
I got a 475MHz stable P5W DH Deluxe without any mods ( stock, rev. 1.03g, 1.85Vmch ).

You did better than I did the best I could do with my mods was 450 bootable and 465 with clockgen.

I still have it collecting dust in my basement...dont ask :(

In all honesty I love my striker for Dual Core but not for Quad.

[fhpchris]

re-run the 680i @ 600 mhz on ram :P

[BenchZowner]

re-run the 680i @ 600 mhz on ram :P

Will do so, when...the Fed-Ex guys decide to push their speed-limits & deliver my long awaited package :P

[Aljechin]

Will do so, when...the Fed-Ex guys decide to push their speed-limits & deliver my long awaited package :P

Nice comparision Bench, don't let you get distracted by all the
BUT BUT BUT-Lamers. Its a clock-to-clock-comparision,
everything else is another slice of bread.

[Ten]

Thx for this comparison, last week i went back to my 975x chips and im happy so far:)

[Sir Lancelot]

Your effort to make this comparison is something to be thankfull for, but I don't think that it will help anybody.

To be a real comparison between both chipsets, you've should compare the best that the 975 can do and the best thew 680i can do.

I don't see the objective of comparing MHz-by-Mhz at 350MHz FSB, when most of the 975 can do 400MHz and the 680i can do more than 450...

After reading your test, my doubt is still the same: at the end, which boards are more powerfull? The 975 ones or the 680i ones?...




Hasta!

[STaRGaZeR]

Your effort to make this comparison is something to be thankfull for, but I don't think that it will help anybody.

To be a real comparison between both chipsets, you've should compare the best that the 975 can do and the best thew 680i can do.

I don't see the objective of comparing MHz-by-Mhz at 350MHz FSB, when most of the 975 can do 400MHz and the 680i can do more than 450...

After reading your test, my doubt is still the same: at the end, which boards are more powerfull? The 975 ones or the 680i ones?...




Hasta!

CLOCK-PER-CLOCK PERFOMANCE means anything to you?

[RealTelstar]

The bench is fair, he is comparing both chipsets doing 2T.

But i second the request of Sir Lancelot. But do it at over 400mhz of FSB. For istance I dunno if the nv680i increases latencies like the i965.

[Sir Lancelot]

CLOCK-PER-CLOCK PERFOMANCE means anything to you?

Yes it does...and it would make sense if both chipsets maxed out at the same fsb...


ARE YOU BLIND?...cos I'm not...

[Big_Sam]

I doubt only of correctness of result in 03Mark.
680i should be much more slowly in this test.
I think that it was necessary to take weaker videocard or faster CPU.

[STaRGaZeR]

Yes it does...and it would make sense if both chipsets maxed out at the same fsb...


ARE YOU BLIND?[/SIZE]...cos I'm not...

NOT? Sure? As already said in this thread, this is a comparison with the SAME settings and hardware. If you want to flame, go and open one called "Tell me the best possible overclock of 975 and 680i". And there, people will say you that it depends of lucky in most cases. Mi BX2 at 450FSB is rock stable, 480FSB bencheable. So don´t come here to cry...

[Sir Lancelot]

NOT? Sure? As already said in this thread, this is a comparison with the SAME settings and hardware. If you want to flame, go and open one called "Tell me the best possible overclock of 975 and 680i". And there, people will say you that it depends of lucky in most cases. Mi BX2 at 450FSB is rock stable, 480FSB bencheable. So don´t come here to cry...


Hey hey hey...I didn't flame any thing...you're the one who started talking with the biggest letters possible!

I never said that the bench was unfair...not precise...or meaningless...got it? (maybe i need bigger font...:stick: :stick: )

What I just said...and I believe it's not stupid, as much as you would like it to be, is that comparing MHz-by-MHz does not mean that 975 is faster, as they limit much earlier than 680i.

I also have a 975x based mobo that can do 450MHz stable, and I didn't came here bragging about it!

So, if asking for a comparison between a 975 @ 420MHz FSB and a 680i @ 500MHz is stupid in your understanding...you better sell your PC and buy a laptop, because it seems that you're a little bit out of the game...


Again, my objective is not flamming...is just to have a better undersdtanding of the systems out there, cos the only one I know about is 975x and I'm curious about other systems as well...


P.S- Is all that flame conversation just because I'm portuguese and you're spanish? Cos if it is...this is not the right place to bring your historical frustrations...And because of that, I wont feed this thread any longer, as I explained myself very well and there's nothing more I can add...

[STaRGaZeR]

P.S- Is all that flame conversation just because I'm portuguese and you're spanish? Cos if it is...this is not the right place to bring your historical frustrations...And because of that, I wont feed this thread any longer, as I explained myself very well and there's nothing more I can add...

LOL WTF man, stop saying things like this... all BS.

It seems that you already know where 975 tops, and 680i too. So dont come here to say that you want want a comparison with the max settings allowed by both chipsets, because:

1. The title of the thread is clear, clock per clock comparison.
2. BenchZowner have said that he will do that test you want when he has all the hardware needed to do it. Is written in his first post, so you must have read it.

The best of the 680i (with a little higher FSB than 975) is the memory overcloking. fhpchris and many others in XS have sucesfully done 600+ on the memory, you only have to search for it.

And the thing about you are portuguese and im spanish... well, its sounds like you have a pain in the ass. Better said, is your historical frustation. Many of you portuguese when see spanish members in a forum, game, etc. start saying stupid things about the history :rolleyes: :D

[fhpchris]

CLOCK-PER-CLOCK PERFOMANCE means anything to you?

If you can even get 975x to run 600 mhz on ram, then do that also. :)
Clock for clock performance only really means CPU, not same on memmory, because both platforms do not really do the same things...

But I can run 600 mhz cas 5-5-5 on 680i, on 2.2v...


People on 975x run ~450-470 cas 3-3-3 (8*450 1:1 = 3600)
People on 680i run ~600-630 cas 5-3-3 or 4-3-3(8*450 3:4 divider = 3600)

It would be nice to compare which is faster in those cases :)

Because you have an XE, just use the multi of 8x ;) (you might need ~2.6v for rams to do 450 though if you have a bad set...:()

[BenchZowner]

Like I've previously mentioned in my OP, I will run another ( maybe 2 ) comparisons between these boards ( P5W DH Deluxe & Striker Extreme ) this week ( weekend probably :D ).
I didn't have the "right" memory kit to push both boards to their limits ( my GeIL kit died doing 1058MHz 3-3-3-8 on the P5W :( ), but I'll be getting some new memory kits this week ;)

I'd like to see where my best ram kit will stop on the P5W & where on the striker...

[sdumper]

Time to jump in :)

The thread is what it is but as I mentioned before the Striker for me has been able to OC better and obtain better SP simply because it can OC higher.

I did 4.715 with relative ease on my Striker and with the same chip I caped out at 4.5 with a heavily volted moded P5W Deluxe....CPUz for P5W DH and e6700 NOT shown in my sig but Striker is...

[dinos22]

If you can even get 975x to run 600 mhz on ram, then do that also. :)
Clock for clock performance only really means CPU, not same on memmory, because both platforms do not really do the same things...

But I can run 600 mhz cas 5-5-5 on 680i, on 2.2v...


People on 975x run ~450-470 cas 3-3-3 (8*450 1:1 = 3600)
People on 680i run ~600-630 cas 5-3-3 or 4-3-3(8*450 3:4 divider = 3600)

It would be nice to compare which is faster in those cases :)

Because you have an XE, just use the multi of 8x ;) (you might need ~2.6v for rams to do 450 though if you have a bad set...:()
975X would still win that SPI contest :)

[fhpchris]

Like I've previously mentioned in my OP, I will run another ( maybe 2 ) comparisons between these boards ( P5W DH Deluxe & Striker Extreme ) this week ( weekend probably :D ).
I didn't have the "right" memory kit to push both boards to their limits ( my GeIL kit died doing 1058MHz 3-3-3-8 on the P5W :( ), but I'll be getting some new memory kits this week ;)

I'd like to see where my best ram kit will stop on the P5W & where on the striker...

Oh, which ram kits did you "invest" in :)

[dinos22]

CLOCK-PER-CLOCK PERFOMANCE means anything to you?

LOL hehehe

[BenchZowner]

Oh, which ram kits did you "invest" in :)

Actually the kits "invested" on me :D
The OCZ Reaper HPC (PC8500 or PC9600 / don't know which one :P ), a Corsair 9136C5 Dominator kit, CellShock PC8000 4-4-4, and a 2GB Fatbody kit :)

If I'm lucky one of these kits will do good ( hopefully :D )

[BenchZowner]

Let's revive this thread to add the second part :D

Part 2:

=======================
Test System
=======================
Motherboard: Asus Striker Extreme ( 680i ) & Asus P5W DH Deluxe ( 975X )
CPU: Intel Core 2 Duo Extreme X6800 ( @ 3.66GHz / 333MHz x 11 )
RAM: Corsair Dominator 9136C5D ( TWIN2X2048-9136C5D )
VGA: Twintech GeForce 8800GTX
HDD: Maxtor MaXLine III 300GB

=======================
Configurations
=======================
FSB: 333MHz ( QDR-1333 )
DRAM/FSB: 1:1
RAM Mode: Linked
RAM Sync: Synchronized
RAM Timings: 3-3-3-4
http://www.benchzone.com/Benchmarking/memset680i-975x.jpg
HyperPath 3 - GPUex - LinkBoost: Disabled
Command Rate: 680i = 1T , 975X = 2T
BIOS Versions: Striker1002 - P5W1601
OS:Windows XP Service Pack 2
Drivers:Intel Chipset Software Installation Utility v8,1,1,1010 - nVIDIA nFORCE v9.53 - nVIDIA ForceWare v97.94

===============================
The results
===============================
Hexus PiFast:
680i = 26.47s
975X = 25.92s

SuperPi 8M:
680i = 181.563s http://www.benchzone.com/datas/thumbs/0-striker-extreme-overclocked-superpi8m.jpg (http://www.benchzone.com/datas/users/0-striker-extreme-overclocked-superpi8m.jpg)
975X = 179.422s http://www.benchzone.com/datas/thumbs/0-p5w-overclocked-superpi8m.jpg (http://www.benchzone.com/datas/users/0-p5w-overclocked-superpi8m.jpg)

Everest Ultimate v400:
680i = Memory Read: 9511MB/s - Memory Latency: 52.8ns
975X = Memory Read: 8554MB/s - Memory Latency: 50.0ns

WinRAR v3,70 Beta 5 Benchmark:
680i = 1599kb/s
975X = 1710kb/s

3D Mark2001SE:
680i = 60824
975X = 61155

AquaMark 3:
680i = 190977
975X = 193073

x3:Reunion Benchmark:
680i = 85.328fps average
975X = 85.731fps average

Windows Media Encoder 9 - HD ( 5MBPS ) Video Encoding Test:
680i = 8min 12s
975X = 8min 2s

[illuminati-hwt]

Good tests. both part 1 and 2.. :-)

although,

Setting the tRAS to 4 isnt smart. The timing has to complete two cycles because it ends/closes too fast. This might be screwing with the results.
tRAS should in this case be minimum 6. Sry to say it :-/
tRAS should not be lower than tCL+tRCD.

The low tRC on 680i might be hurting performance too. Not sure though. Dont let tRC lower than tRAS+tRP.


edit.. lower than tCL+tRCD

[BenchZowner]

Setting the tRAS to 4 isnt smart. The timing has to complete two cycles because it ends/closes too fast. This might be screwing with the results.
tRAS should in this case be minimum 6. Sry to say it :-/
tRAS should not be lower than tCL+tRCD.

It's not hurting performance :)

There's a 3rd part covering more stuff :D

And fast close doesn't occur every time ;)


The low tRC on 680i might be hurting performance too. Not sure though. Dont let tRC lower than tRAS+tRP.

tRC = tRAS + tRP

[Eastcoasthandle]

Thanks for the re-fresher on this issue. Maybe some of those 680i users will finally accept their fate (something I saw illustrated in the past). Oh and let me be the first to post the infamous, "My 680i does this benchmark better <fill in false numbers here> "

[SMa]

in your signature it says you have a P5B Deluxe?
which has a P965-chipset...
would be great if you added that board to your tests :p:

nice comparison though! :clap:

[BenchZowner]

in your signature it says you have a P5B Deluxe?
which has a P965-chipset...
would be great if you added that board to your tests :p:

nice comparison though! :clap:

I will be adding the P5B in the game in about 2 weeks ( I really need to finish some reviews & round-ups right now ;) )

[Eastcoasthandle]

LOL not one comment from the 680i camp muhahaha!!!
Maybe a nice link post over from their forum perhaps?

[CedricFP]

Wow, great thread! This is really useful. Good going :D

[Eastcoasthandle]

Wow, still no posts, where is the love?

[Console]

Nice tests , Going to pull the triger on a c2d system soon and glad to have info like to to use as a reference