Go Back   nV News Forums > Hardware Forums > Benchmarking And Overclocking

Newegg Daily Deals

Reply
 
Thread Tools
Old 10-04-05, 05:53 PM   #133
Clevor
Registered User
 
Join Date: Sep 2002
Posts: 220
Default Re: Official AquaMark 3 thread

Orbitech, what about the mem controllers on X2s? Rumor is they're not the best. Could be the deal where you have 2x the chance of getting a bad one. You say you are running 9/10 divider on UTT??? Is that 9/10 at 270x10? You need very good memory performance for benching, stuff like near 270, 2-2-2-5 with UTT or TCCD at 300, 2.5-3-3-7.

My theory is the best cores are put in FXs and charged a premium. The next best ones go in 3700-4000+ San Diegos or maybe Venices, which is why they are usually good for 2850-2950 on water. The rest go in the X2s as they are not gonna waste two good cores in them. AMD figures most guys using them are into multitasking and not overclocking. There's a guy on XS who tried seven X2s before he found one that did 3000 on phase at 1.58 volts.

If comparing to Intel, figure the price/performance ratio. I mean a 4400+ is $530 right now.

There are a lot of contrary things I see with overclocking and such. I'm playing with my ole P4T533-R right now (850E chipset). Never seen anything boot WinXP faster than RDRAM boards. I mean they boot as fast or faster than an FX at 3000 mhz. They exit damn fast too. And with 2x256 MB ram.

Yeah, I'm hardly an Intel fanboy. Goes back to the day my 300 PentiumII died and killed a $360 Asus board with on-board Ultra160. Intel refused to reimburse me for the board. As much profit as they make . . .

I'm mainly into seeing what a single video card at default clock can do. It's pretty easy to get earth shattering scores with a highly overclocked card or SLI. Aquamark is an old game and it doesn't support dual core. There is no reason dual cores should provide a performance benefit. Moreover, if you run where most people game these days: 1600x1200x32, 4xFSAA, 16xAniso, that is all video card. A monster CPU or dual cores won't help you there.
__________________
DFI SLI-D
510-2 BIOS
FX-57, 0530APMW
300x10 at default VCORE (1.36 volts)
2x512MB GSkill FF at 2.5-3-3-7 (2.8 VDIMM)
X850XT PE at default clock
Corsair HC200

SuperPI 32M=24'14"
SuperPI 1M=27.187"
Clevor is offline   Reply With Quote
Old 10-04-05, 06:08 PM   #134
SH64
MAXIMUM TECH
 
SH64's Avatar
 
Join Date: Jul 2003
Location: Indiana
Posts: 12,202
Default Re: Official AquaMark 3 thread

Clevor but some people are getting performance increase with the multi-threaded drivers even at 1600x1200,4xAA,8xAF ... see here for exmaple :
http://www.nvnews.net/vbulletin/show...&postcount=156

you can find more at the drivers section .. the feedback of the 81.82 drivers thread.
__________________


- "My name is RAM and my tank is full"

http://warhawk64nv.mybrute.com/ <-- pupils go thaarrr! Or,
http://silenthunter64.mybrute.com
SH64 is offline   Reply With Quote
Old 10-04-05, 06:23 PM   #135
Clevor
Registered User
 
Join Date: Sep 2002
Posts: 220
Default Re: Official AquaMark 3 thread

Quote:
Originally Posted by SH64
Clevor but some people are getting performance increase with the multi-threaded drivers even at 1600x1200,4xAA,8xAF ... see here for exmaple :
http://www.nvnews.net/vbulletin/show...&postcount=156

you can find more at the drivers section .. the feedback of the 81.82 drivers thread.
It's possible the old drivers were not optimized for dual core and maybe were interfering with performance, i.e. you will see a performance diff now if you have a dual core CPU.

If performance is indeed better against a single core, same clock, perhaps it's because you can use one CPU for the bench and the other to handle the background TSRs???
__________________
DFI SLI-D
510-2 BIOS
FX-57, 0530APMW
300x10 at default VCORE (1.36 volts)
2x512MB GSkill FF at 2.5-3-3-7 (2.8 VDIMM)
X850XT PE at default clock
Corsair HC200

SuperPI 32M=24'14"
SuperPI 1M=27.187"
Clevor is offline   Reply With Quote
Old 10-04-05, 06:28 PM   #136
orbitech
Registered User
 
Join Date: Sep 2005
Posts: 20
Default Re: Official AquaMark 3 thread

Quote:
Originally Posted by Clevor
Orbitech, what about the mem controllers on X2s? Rumor is they're not the best. Could be the deal where you have 2x the chance of getting a bad one. You say you are running 9/10 divider on UTT??? Is that 9/10 at 270x10? You need very good memory performance for benching, stuff like near 270, 2-2-2-5 with UTT or TCCD at 300, 2.5-3-3-7.
That seems to be true and everyday I believe it even more and more. Definately mem controllers are not the shiny point of dual cores .. Well even if I wanted I couldn't do 270 at 1:1 with 2-2-2-5 with this ram.But at least the ram I use has been tested on an excellent mem controller and it can do 256 at 2-2-2-5. I cannot do more than 232 at these timings with my Toledo.

Quote:
If comparing to Intel, figure the price/performance ratio. I mean a 4400+ is $530 right now.

There are a lot of contrary things I see with overclocking and such. I'm playing with my ole P4T533-R right now (850E chipset). Never seen anything boot WinXP faster than RDRAM boards. I mean they boot as fast or faster than an FX at 3000 mhz. They exit damn fast too. And with 2x256 MB ram.
Compare them in what? A fair price/perf comparison is with Intel's dualies and there x2 3800+ destroys the competition. If all you do is just multitasking and caring for speed in Windows then Intel's single core is just fine. But if you're talking about REAL world demanding applications performance AND you want to multitask then I don't see Intel anywhere near AMD dual core...
Try render something in 3dsmax and do something else demanding with your single core Intel... Then come tell me about Intel's H/T.

Quote:
I'm mainly into seeing what a single video card at default clock can do. It's pretty easy to get earth shattering scores with a highly overclocked card or SLI. Aquamark is an old game and it doesn't support dual core. There is no reason dual cores should provide a performance benefit. Moreover, if you run where most people game these days: 1600x1200x32, 4xFSAA, 16xAniso, that is all video card. A monster CPU or dual cores won't help you there.
Well I think you're really wrong about Aquamark3. It has a great engine and is still one of the most reliable (if not the best) engines to count total pc gaming performance. Even if you change memory timings you'll notice good differences in it. Of course is not optimized for dual cores , but as I said immediately showed the gains from multi threaded drivers.. And of course is influenced by SLI or o/ced gpus BUT SURELY no more than in real gaming situations... CPU also is a major component of it, so what's your problem with it? I think it has a very remarkable engine. Counting everything in your system (cpu,ram,gpu) as it should. Not like stupid synthetic benches that are made to show your gpu e-penis and no matter what cpu you have scores are comparable if you have the same gpu. LOL. I can't remember another benchmark that is so "complete" as Aquamark3 is. 3dmark01 used to be one but not anymore...
__________________
A64 x2 4400+ @2.7GHz (xp90c ) (10X270)
DFI LanParty SLI - DR (bios v 623-3)
2x512 Mushkin Redline 3500 (2-2-2-5) 9:10
MSI GeForce 7800GTX
Antec Neopower 480W
Coolermaster Stacker
Audigy 2 ZS @ Logitech Z-680
Leadtek Winfast PVR2000
WD 36.7 SATA Raptor
2x80 WD 7200rpm
Plextor 708A , NEC 3520A
Aerocool Gatewatch Panel
HP printer scanner
orbitech is offline   Reply With Quote
Old 10-04-05, 06:36 PM   #137
orbitech
Registered User
 
Join Date: Sep 2005
Posts: 20
Default Re: Official AquaMark 3 thread

Quote:
Originally Posted by Clevor
It's possible the old drivers were not optimized for dual core and maybe were interfering with performance, i.e. you will see a performance diff now if you have a dual core CPU.

If performance is indeed better against a single core, same clock, perhaps it's because you can use one CPU for the bench and the other to handle the background TSRs???
Exactly my point..Well of course 7x.xx were not optimized for dual cores. Nvidia has started experiments with multi threaded drivers in the 8x.xx drivers. Beyond game increases with them, let's take my Aquamark3 perf with 81.82.

78.01 : 91-92fps
81.82 : 105-106fps !!!
Pretty good boost in performance isn't it? (That's why I asked for a FX-57 with GTX score to compare it)

As I said what these drivers do is assigning vertex offload to the second core of my dualie leaving the gpu with more processing power.

P.S : Scores with these drivers and dual cores compared to a single core using same speed,ram timings(in my case let's say an A64@2.7 using the same timings in ram) are not only less but not comparable.
Anyway even w/o these drivers dual cores have better gaming performance at same speeds with a single core. Imagine with them
__________________
A64 x2 4400+ @2.7GHz (xp90c ) (10X270)
DFI LanParty SLI - DR (bios v 623-3)
2x512 Mushkin Redline 3500 (2-2-2-5) 9:10
MSI GeForce 7800GTX
Antec Neopower 480W
Coolermaster Stacker
Audigy 2 ZS @ Logitech Z-680
Leadtek Winfast PVR2000
WD 36.7 SATA Raptor
2x80 WD 7200rpm
Plextor 708A , NEC 3520A
Aerocool Gatewatch Panel
HP printer scanner
orbitech is offline   Reply With Quote
Old 10-04-05, 06:58 PM   #138
Clevor
Registered User
 
Join Date: Sep 2002
Posts: 220
Default Re: Official AquaMark 3 thread

As far as ram is concerned, it's all in the sticks. Unless you work for a mem manufacturer and can cherry pick sticks, it's all luck.

I don't believe this mem controller business, but I guess I was lucky and all four Winnies (including a 3000+) and both the SDs I have now just so happen to have great mem controllers. I have two sets of TCCD that can do 300+, 2.5-3-3-7 and I can't even gain one more mhz on the FX-57 or an FX-55 I tested couple of months ago. ??? I was able to run 300, 2.5-3-3-7 even on the 3000+ Winnie I had. So much for mem controllers.

Noooo, I'm not that much of an Aquamark fanboy. I like a lotta benchmarks. Fivemark minimum: all four Marks and Aquamark. I'm one of the few people who still rely on 3DMark2000 for benching and stability testing. All these are run at 1028x764 and so are CPU dependent. 3DMark2003/2005 don't move much with the CPU, but I use them as stability tests. Interestingly, I get higher 3DMark2000 scores on a S754 Clawhammer than on a S939 Winnie. It's in the 1 MB L2.

Aquamark is a very good quick-and-dirty test for ram stability. I don't multitask or video encode or anything. I don't even game actually, no time. My friend has a dual Xeon rig with 4 RAID drives and on another, uses a 3.0C clocked to 3.6 for burning DVDs. No complaints from him. Dual cores are for guys who want the latest and the greatest. At least HT was virtual and didn't affect the overclock on the CPU.

I'd like to get an X2 eventually, but they have to improve the cores. 2900 mhz is minimum I'd accept for that kind of prices. I'd then provide some objective benchmark comparisons with single core.
__________________
DFI SLI-D
510-2 BIOS
FX-57, 0530APMW
300x10 at default VCORE (1.36 volts)
2x512MB GSkill FF at 2.5-3-3-7 (2.8 VDIMM)
X850XT PE at default clock
Corsair HC200

SuperPI 32M=24'14"
SuperPI 1M=27.187"
Clevor is offline   Reply With Quote
Old 10-04-05, 07:02 PM   #139
Clevor
Registered User
 
Join Date: Sep 2002
Posts: 220
Default Re: Official AquaMark 3 thread

Comparing at X2 at heh, 2700 mhz to a single core at 2700 mhz is not a fair and realistic comparison. You have to assume the average person can get a better overclock with the single core (San Diego anyway).

I mean even my San Diegos do 2900-2980 on water, at around 1.568-1.60. If you run CPU dependent benchmarks, no way the smaller CPU, even dual core, will prevail.
__________________
DFI SLI-D
510-2 BIOS
FX-57, 0530APMW
300x10 at default VCORE (1.36 volts)
2x512MB GSkill FF at 2.5-3-3-7 (2.8 VDIMM)
X850XT PE at default clock
Corsair HC200

SuperPI 32M=24'14"
SuperPI 1M=27.187"
Clevor is offline   Reply With Quote
Old 10-04-05, 07:27 PM   #140
orbitech
Registered User
 
Join Date: Sep 2005
Posts: 20
Default Re: Official AquaMark 3 thread

Quote:
Originally Posted by Clevor
Comparing at X2 at heh, 2700 mhz to a single core at 2700 mhz is not a fair and realistic comparison. You have to assume the average person can get a better overclock with the single core (San Diego anyway).

I mean even my San Diegos do 2900-2980 on water, at around 1.568-1.60. If you run CPU dependent benchmarks, no way the smaller CPU, even dual core, will prevail.
Never say never

I'm telling this again. Dual cores prevail in gaming performance compared to an equivalent in speed single core. And many guys (including me ) have reached 2.8 with 1.55-1.65 vcore or watercooling. Now add to this the huge boost from the multithreaded drivers (avg 13% for me) and where does this lead you? Do you think a 2.9-3.0 single core can give 13% more gaming performance than a 2.7-2.8 dual core?? NO FUxxx WAY ... So dualie gaming performance (with these drivers) @2.8 is more like comparable to the gaming performance of an oc/ed A64 @3.1-3.2 ??
__________________
A64 x2 4400+ @2.7GHz (xp90c ) (10X270)
DFI LanParty SLI - DR (bios v 623-3)
2x512 Mushkin Redline 3500 (2-2-2-5) 9:10
MSI GeForce 7800GTX
Antec Neopower 480W
Coolermaster Stacker
Audigy 2 ZS @ Logitech Z-680
Leadtek Winfast PVR2000
WD 36.7 SATA Raptor
2x80 WD 7200rpm
Plextor 708A , NEC 3520A
Aerocool Gatewatch Panel
HP printer scanner
orbitech is offline   Reply With Quote

Old 10-04-05, 07:51 PM   #141
Clevor
Registered User
 
Join Date: Sep 2002
Posts: 220
Default Re: Official AquaMark 3 thread

Quote:
Originally Posted by orbitech
Never say never

I'm telling this again. Dual cores prevail in gaming performance compared to an equivalent in speed single core. And many guys (including me ) have reached 2.8 with 1.55-1.65 vcore or watercooling. Now add to this the huge boost from the multithreaded drivers (avg 13% for me) and where does this lead you? Do you think a 2.9-3.0 single core can give 13% more gaming performance than a 2.7-2.8 dual core?? NO FUxxx WAY ... So dualie gaming performance (with these drivers) @2.8 is more like comparable to the gaming performance of an oc/ed A64 @3.1-3.2 ??
Have you tried to sell your point on XS, that a dual core at 2.8 will outperform a San Diego/FX at 3.1-3.2 gig, in gaming at that? Hmmm, FiveMark too? Why not throw in benching in general? I'd like to see a comparison in SuperPI 32M scores.

If you also assume the single core will overclock higher and have a better mem controller those faster CAS timings will make a BIG impact on performance. Since the cores on the X2s are not the best naturally they will run hotter and the mem controllers may be poorer.
__________________
DFI SLI-D
510-2 BIOS
FX-57, 0530APMW
300x10 at default VCORE (1.36 volts)
2x512MB GSkill FF at 2.5-3-3-7 (2.8 VDIMM)
X850XT PE at default clock
Corsair HC200

SuperPI 32M=24'14"
SuperPI 1M=27.187"
Clevor is offline   Reply With Quote
Old 10-04-05, 09:20 PM   #142
orbitech
Registered User
 
Join Date: Sep 2005
Posts: 20
Default Re: Official AquaMark 3 thread

Quote:
Originally Posted by Clevor
Have you tried to sell your point on XS, that a dual core at 2.8 will outperform a San Diego/FX at 3.1-3.2 gig, in gaming at that? Hmmm, FiveMark too? Why not throw in benching in general? I'd like to see a comparison in SuperPI 32M scores.

If you also assume the single core will overclock higher and have a better mem controller those faster CAS timings will make a BIG impact on performance. Since the cores on the X2s are not the best naturally they will run hotter and the mem controllers may be poorer.
Do not put ram for the time being in our example. Not every dual core has bad mem controllers and I've seen ppl running at blazing ram speeds with a dualie.Now of course I don't assume a 2.8Ghz dual core will outperform a 3.1-3.2Ghz single core. If you read my post better you'll see that. What I say is that it will outperform a 2.9-3.0Ghz single core with and have in many instances comparable gaming performance with a 3.1-3.2 with these drivers (other things equal mem etc.).
But we'll never now until I can find some game benches from an o/ced FX-57@3-3.1GHz and a 7800gtx.That's why I asked you if you want to ran some game benches to compare results with mine(FC,DOOM3,FEAR,SC:CT etc). I'm telling you ,you seem to underestimate these drivers... 10-15% gain in FPS in many cases, is a lot of processing power for a cpu How much do u think can a 3.1-3.2Ghz differ in performance from a 2.8 dual with these drievrs(other things equal)? Can it surpass that percentage(10-15%) easily you think? I don't.

As for SuperPI 32M well of course with better ram and your O/CED FX-57 you will win me. It's a single thread application, so what's the point here? For reference I score around 28.5secs in 1M with @2.8GHz and ram @9:10 (2-2-2-5). Not bad..
Do you want to try Crystalmark, Sandra which support dual core and ran some ALU or multimedia tests to tell me? We need to compare aples with apples to have useful feedback.
__________________
A64 x2 4400+ @2.7GHz (xp90c ) (10X270)
DFI LanParty SLI - DR (bios v 623-3)
2x512 Mushkin Redline 3500 (2-2-2-5) 9:10
MSI GeForce 7800GTX
Antec Neopower 480W
Coolermaster Stacker
Audigy 2 ZS @ Logitech Z-680
Leadtek Winfast PVR2000
WD 36.7 SATA Raptor
2x80 WD 7200rpm
Plextor 708A , NEC 3520A
Aerocool Gatewatch Panel
HP printer scanner
orbitech is offline   Reply With Quote
Old 10-04-05, 10:04 PM   #143
orbitech
Registered User
 
Join Date: Sep 2005
Posts: 20
Default Re: Official AquaMark 3 thread

106.469 with 81.84 (other things equal)

__________________
A64 x2 4400+ @2.7GHz (xp90c ) (10X270)
DFI LanParty SLI - DR (bios v 623-3)
2x512 Mushkin Redline 3500 (2-2-2-5) 9:10
MSI GeForce 7800GTX
Antec Neopower 480W
Coolermaster Stacker
Audigy 2 ZS @ Logitech Z-680
Leadtek Winfast PVR2000
WD 36.7 SATA Raptor
2x80 WD 7200rpm
Plextor 708A , NEC 3520A
Aerocool Gatewatch Panel
HP printer scanner
orbitech is offline   Reply With Quote
Old 10-04-05, 11:14 PM   #144
Clevor
Registered User
 
Join Date: Sep 2002
Posts: 220
Default Re: Official AquaMark 3 thread

"As for SuperPI 32M well of course with better ram and your O/CED FX-57 you will win me. It's a single thread application, so what's the point here? "

What's the point??? I dunno, why do we guys run SuperPI 32M over at XS? Since it's single threaded it's useless I guess. So name a bunch of existing dual threaded apps and 3D games. Yes, you'll win for sure when those come out!

I am at 300x10, 2.5-3-3-7 at 1.36 VCORE, for starters. Yes, we are indeed comparing apples with oranges.

If you really think your dual core is superior, post something over at XS and let's see what happens. I was just over there and I see nothing about these Nvidia drivers setting the world on fire. Not to say you are wrong.
__________________
DFI SLI-D
510-2 BIOS
FX-57, 0530APMW
300x10 at default VCORE (1.36 volts)
2x512MB GSkill FF at 2.5-3-3-7 (2.8 VDIMM)
X850XT PE at default clock
Corsair HC200

SuperPI 32M=24'14"
SuperPI 1M=27.187"
Clevor is offline   Reply With Quote
Reply


Thread Tools

Posting Rules
You may not post new threads
You may not post replies
You may not post attachments
You may not edit your posts

BB code is On
Smilies are On
[IMG] code is On
HTML code is Off
Forum Jump


Similar Threads
Thread Thread Starter Forum Replies Last Post
Official GeForce GTX 670 Review and Discussion Thread MikeC NVIDIA GeForce 600 Series 23 10-28-12 10:19 AM
Official nForce 2 Thread Matthyahuw CPUs, Motherboards And Memory 148 04-16-03 04:39 AM
Official Linux driver 1.0-3123 thread bammbamm808 NVIDIA Linux 126 12-05-02 06:21 PM
Official Grand Theft Auto 3 thread volt Gaming Central 6 08-01-02 07:04 AM

All times are GMT -5. The time now is 04:09 AM.


Powered by vBulletin® Version 3.7.1
Copyright ©2000 - 2014, Jelsoft Enterprises Ltd.
Copyright 1998 - 2014, nV News.