W1nTry Benchies

[W1nTry]

Well I have rebuilt my system somewhat and now I have moved from a K7 2600+ Barton and A7V880 to a A64 3500+ and a Epox nf3 250Ultra Mobo. The other specs are the same but i'll put there for good measure (shame I don't have the benchies from the first rig anymore):

A64 3500+ @ 2.2GHz (stock)
1GB Kingston HyperX @ 333MHz (will work on this when I get money, OCZ 512x2 Gold PC3200)
Epox EP-9NDA3J Socket 939 NVIDIA nForce3 Ultra
ATI 9800Pro 128MB
2x80GB Hitachi SATA 7200RPM
Creative SoundBlaster Audigy

CPU-Z, PCMark04, CrystalMark, Super Pi



3DMark01/03/05

[W1nTry]

There is an adjustment to my rig. I have new memory: Corsair TWIN XMS (2-2-2-5-1T) 2x512MB
Also the rig is running OCed to 2.42GHz. Everything else remains as before.

CPU-Z, Everest, PCMark04, CrystalMark, Super Pi



3DMark01/03/05


Quake 3 (High settings), Doom 3 (Med. setting) @ 1024x768 no AA/AF

[W1nTry]

I got it stable @ 2.6GHz
System changes:
X800XT AIW
SB Audigy 2ZS Platinium

3DMark01/03/05

[W1nTry]

Well I promised the OC results and here they are... i'll comment on the OC after you see them, they start with the info stuff CPU-Z, system screenie then onto benchies, have a look see:


You'll notice its 220x11 as opposed to say 240x10 or 250x10. I am not sure why I can't get those settings stable as i've done it before.. it could be a power issue but I have not yet determined it. So that was what I had stable... I didn't run Prime95 24hrs (doh have that kinda time) I did however run it for about 2 hours then ran all the benchies repeatedly (in the 3DMark cases) AND ran the system whole day without a single crash.


Some system benchies... memory, CPU power, etc. Nothing outta the ordinary


Some overall systems tests, the regular.. I do have sciencemark to run still... now remember that one...


The staple 3DMark line up.. I forgot to take a screenie of the OC on the card either with ATI Tool or RivaTuner. However to fill you in the Card was OCed to 530MHz Core and 575MHz Memory (1150MHz effective). The 3DMark benchies were run 3 times (set in setup for each) just to make sure no fluke scores occured.


And last but not least a doom3 timedemo test for more real world performance.

[W1nTry]

Latest change:
CPU 4400+ Toledo core
All the other parts have remained save the addition of a Logitech Z5500 5.1 speaker system.

cpu-z, PCMark05, CrystalMark



3DMark01/03/05/06



1 Thing worth mentioning is that some of these benchmark were run using different drivers and the system was reinstalled recently which would improve performance somewhat. I will update the necessary screenshots as I get them, however this is a good idea of how the system performs.

[steveunderwood0]

talkin' about the z5500's....I saw one company in Trinidad selling those for $3,500 TT...that's just insane to me that anybody would pay that...considering you can get for $1300 new...if you are willing to go and get them from the states.

[W1nTry]

Well you see therein lies the catch underwood. Speakers have an prety HEAVY tax on them... something like 25% or so... plus shipping something of that size and weght isn't cheap and even then it's still 300USD to start with so you start with 1800 then add shipping and a HEAVY tax and you reach well over 2G easily. But even so what does this have to do with my Benchmarks?

[steveunderwood0]

heh.  That's true, about the benchies...just noticed that you were using the z5500's.  How do the speakers affect your benchmarks?  How much do the drivers for yoru soundsystem affect your personal benchmarks?  Have you measured any noticeable difference with these speakers versus your former set? 

[W1nTry]

Well the speakers don't in any way detract from the benchmarks save I was in a db benchmark... but thats completely off this thread. I just mentioned it as its part of my rig to add to the system as a whole. As for the drivers, they were seen to have a SERIOUS difference. Seeing as my card is from the X8xx range of cards and the X1k range has a different memory achtitecture, the only thing I could have surmised is that in the X1k series they optimized the memory controller to boost OpenGL performance, however seeing as my card is a X8xx series and cannot use those memory optimization I saw a HORRIFIC performance DROP in the OpenGL benchmark for CrystalMark. And I mean CATASTROPHIC difference. If you reveiw even the performance of the 9800Pro in my first set of results, it scored well over 30K, with 6.7 Catalyst drivers installed I saw a disturbing 7000 points in the exact same benchmark. FAR less for the X800. I have subsequently cleaned out all remnants of the 6 series drivers from my rig and replaced it with 5.13 Catalyst. I use the older drivers as they were optimized for my series of cards vs. the newer cards. My score as you see in the last set of results is actually higher than the 9800Pro which makes sense when you consider the previous score from the OCed X800XT results was 38K.

[TriniXaeno]

Scary

I know Nvidia was accused of doing something similar a few moons ago.

That is, "optimize" the new drivers and in so doing reduce the performance of older cards. Needless to say this helps give the impression that the new cards are scoring better in benchmarks.

I am sure they have good reason but it all smells of an effort to sell the latest graphics cards. They know that people will always want to run the latest and greatest drivers. Staying with the old detonator/catalyst may leave you without critical fixes for games and the likes.

Nasty business.

awesome thread W1ntry. I am jealous.

[W1nTry]

Scary

I know Nvidia was accused of doing something similar a few moons ago.

That is, "optimize" the new drivers and in so doing reduce the performance of older cards. Needless to say this helps give the impression that the new cards are scoring better in benchmarks.

I am sure they have good reason but it all smells of an effort to sell the latest graphics cards. They know that people will always want to run the latest and greatest drivers. Staying with the old detonator/catalyst may leave you without critical fixes for games and the likes.

Nasty business.

awesome thread W1ntry. I am jealous.

Yeah, I too smell something fishy in the driver dept. there. I mean to take a performance hit of almost 80% is unacceptable... maybe if I post these findings to some review site or the like and they make enough noise about it someone at DAAMIT and Nvidia will notice and comment on these findings. LOL @ jealous, you running 7900GT :p I eh have dat kinda horsepower YET....

[Beomagi]

What cooling are you using and what's the location?  T'dad?

Fun to track the updates and changes. Love the smell of fresh hardware

was telling my lil sis about my pc when she saw tubes running through it. explained our first pc was 40MHz lol. 40 - hard to believe they started in the kHz. I remember back in the comp lab in pres machines were 4-8MHz with monochrome amber screens. then they jumped to 133MHz and win 95. Havent been back there in ages. Wonder if they're still running that.

It's funny that that happenned - usually there's just a conditional check on the core version to determine if to use the new code or not. Bad foul up.

[W1nTry]

As of now I am using stock cooling that comes with the X2, which for all intents and purposes is pretty decent. It's no Nautilus but it suits me ftw. As for the environment, usually in AC at roughly 75-79. I have run into an unfortunate problem that I think is power related. I am working on getting a single 12V rail Enermax 550W PSU to replace the rather generic one presently in my case. My Thermaltake 480W just wasn't cutting it with teh X2, damn thing even failed to boot imagine that... since I have been forced to revert to another video card (ala 9600) until I get a more worthy PSU. Failing that it is the mobo to blame. I am weighing meh options at present seeing as I should just go PCIe instead... but time will tell. I not sweating on meh rig lately since I get Live! so I not really using all the video power to begin with. Will update when time comes. Oh btw Beo, post some scores man.

[W1nTry]

Alright, the system is stable again (still using a generic PSU unfortunately and meh Audigy is still outside) however all seems to be well up and running. I have not OCed the CPU (citing power issues that put me here in the first place) however that has not stopped me from pushing the video. These are updated benchies (save and except SiSandra which will be added soon). Sure they're not stella but not half bad. Usual round up of benchies and info screenshots.

CPU-Z
 

ATITool - a modest 6.7% OC on Core and 9.8% OC on memory. I am sure I can hit slightly higher, but I always rather safe than sorry. Still provides measureable difference to the previous results though.


CrystalMark, ScienceMark, Aquamark


3DMark 01/03/05/06

[W1nTry]

Well folks, it seems that I have a regular habit of upgrades... so its that time again. I've not done a complete overhaul in years and that has not changed. I have made 3 changes this time: mobo, video, PSU. Let me give the low down on my current system:

Athlon64 X2 4400+ (2.2GHz stock)
1GB Corsair XMS LL (512MBx2 2-2-2-5-1T) PC3200 (at STOCK freq 400MHz)
Asus A8N-E (nforce 4 chipset, no SLI)
Connect3D X1900XT 256MB version (STOCK clocks: core 621MHz mem 720MHz - note these should be 625 and 725 according to spec, but they not)
Thermaltake Soprano case
OCZ 700W SLI cert PSU
Creative Audigy 2ZS Platinum
NEC DVD burner 3500A or something so (standard stuff)
Hitachi 80GB SATA150 HD (8MB) + Maxtor 200GB SATA150 (8MB)

Note the system is ALL stock, no OCing whatsoever thus and all apps installed (AV, antispam, office, etc). So let's get to the nitty gritty benchies:

CPU-Z cpu/mem/mobo


3DMark 01/03/05/06 - ALL benches run @ res 1280x1024, NO AA, 4x AF (save 3DMark01)


CrystalMark/PCMark05/ScienceMark/SuperPI


It's interesting that 3DMark01/03 show lower performance on the X1900XT than on the X800XT, also the CrystalMark performance is lower. I rule this out to drivers as I am running Catalyst 7.1 at present and have yet to determine if this is the best choice for this card. I say this as it was experienced with my X800XT that drivers > 5.12 showed DRASTIC drops in OpenGL performance (I believe that is mentioned in this thread somewhere). Thus it is possible that a 6 series driver will allow for better performance in synthetic benchmarks than the brand new 7 series catalyst. Either way the system is stock and i'll update my scores when I push it. I have LOTS of power (PSU) so that's not even remotely an issue and also the system runs pretty cool, as low as 32 degrees CPU and 34 mobo @ idle and 45/40 under load. The Soprano case with 2x120mm fans + PSU with 1x120mm fans keep the case very cool... oh and there's a 90mm from the side window over the CPU as well.

[W1nTry]

2.6GHz stable

Desktop shot


CPU-Z


3DMark01(default)/01/03/05/06 (1280x1024 w/ 4x Ansio, no AA)


SuperPI

[W1nTry]

[Eroo]

Love the armour case

[daniboy79]

Love the armour case

Soprano i believe - i was considering this case, but i can't understand how air is being fed to the 120mm front case fan

[W1nTry]

You're quite right, its a Soprano. That aside I think I could help answer your question, if you look at the 4th pic in my post of pics (4th on the top) you will notice the grill in front of the fan. It also has a filter to reduce unwanted dust. As for beyond that with the door, though not ideal in the sense that it will restrict air to some extent, the door is not sealed to the front of the case and air can get in around the entire perimited of the door. By now it should be easy to tell whats inside and my rig and it idles at a cool 32-35 degrees hitting a high of 48 when under full load with NO ac on (Ambient temp in my room can hit 91 degrees farenheit, so all things considered it runs very cool even with relatively high external temps.