New FX-64, "Ship in a bottle"

Show off!!

damn.. ;^)

MM wrote:

Wait a minute guys

This test is affected by graphics resolution.

My time of 27.06 was taken with the quality slider all the way to the right (highest). With the slider all the way to the left I get 20.625. With the "High Quality" button checked I get 21.83.

What does everyone use for a common setting ?

Regards

Mark

FWIW: What a dream compared to my old AMD 1200(89sec.)

50 rebuilds Best 30.9 sec. Worst 32.1

AMD FX-64 NVIDA FX-500 MIRRORING RAID 120 GIG HD

2 GIG DUAL CHANNEL RAM DVD & CD RW

About $2800.00 Box only at XI-Computer.com

Mike Eckstein

You are right

25.03. for 50 Optimized for speed (image looked like hell) Dual 3.2 Ghz Xeon w. 2mb cache Quadro4 980XGL

Worse, 49.5 sec.!!!!!! Everything maxed out! Transparency and optimized edges make a BIG difference.

Mr. Me

message

BTW, Intel rocks! Here comes Nacona!!!!!

Me.

Why do you think these times are so much slower than the dual Opterons?

Jerry Forcier

Michael Eckste>FWIW:

Why do you think these times are so much faster than the Athlon FX 64 - I thought they were basically the same cpu and that SWX didn't benefit from dualys?

Jerry Forcier

MM wrote:

Jerry,

Don't know for sure, but one thing we have to do is make sure everyone's using the same display settings, and SW version/SP. My numbers at work came from SW2003 SP5.1.

After I got home I started experimenting with my old Supermicro dual PIII

1ghz. The fastest times seem to result from setting the display to plain non-accelerated wireframe. I'm assuming this takes display completely out of the picture. Also, at least on my home machine, duals seem to make quite a difference 57 seconds vs 65. I'll have to try some of this at work tomorrow.

I also have a new 64FX-51 system I'm building. Im using a Gigabyte GA-K8NNXP-940, with the same 980XGL. They sent me the wrong HD so it's not completely together yet. When I finish it I'll run some side by side tests.

Regards

Mark

It looks like Mike's defaults are the most torturous: Highest image quality and shaded with edges. Window size is also highly significant. Anti-alias edges seems to be entirely up to the video. I can cut my score by more than a third by cheating.

I'm thinking 1024x768 should be the standard, to include laptop systems. Highest image quality and shaded mode with edges (Mike's defaults) are the most demanding, so they should be kept. If somebody is going to be dishonest and change these settings to make themselves look good, well then they might as well just admit to themselves that they're liars and report phony numbers. Nobody can stop them.

What do you all think?

Time Window size Image quality Display anti-alias edges

80.04 2048x1536 highest shaded w/edges off 78.25 1024x768 highest shaded w/edges off 58.75 2048x1536 lowest shaded w/edges off 56.18 1024x768 lowest shaded w/edges off 75.90 1024x768 highest shaded off 56.10 1024x768 lowest shaded off 78.33 1024x768 highest shaded w/edges on 79.61 1024x768 highest hidden lines off 79.05 1024x768 highest HLR off 60.22 really tiny highest shaded w/edges off 51.94 really tiny lowest shaded w/edges off 51.06 really tiny lowest shaded off

Athlon 1400

512MB CAS2 DDR266 Quadro4 750XGL VIA KT266

Dale,

Seems a bit complicated but It sure covers allot of bases. I don't see a post where Mike (I'm assuming the Mike that started this thread) lists his settings. Is my news server messed up ?

The tests I did at work were at 1280 x 1024 true color (if that means anything). If I get time, I'll run all those combos tomorrow. The screen resolution doesn't seem to affect my home machine much.

Regards

Mark

Wow! What massive differences! It's obvious that, for the numbers to be representative, there has to be agreement as to the settings. I agree with that philosophy, but what about the general requirements to always run LCD's at their native resolution - do you think that only has to do with picture quality, or might it actually give faulty timing results to run a 1920 x 1200 LCD (my Dell 8600 laptop) at 1024 x 768?

Sincerely, Jerry Forcier

Dale Dunn wrote:

I'm anxious to purchase components, but I'll try to wait for your side by sides.

Thanks for the resp>Jerry,

"MM" wrote in news:jbhZb.140$ snipped-for-privacy@newssvr27.news.prodigy.com:

Image quality and shading with edges are settings saved with the file. All the other settings are system settings. So if you just load the model, you have Mike's defaults.

You don't even need to change your desktop resolution. Just resize the SW window. I use an app called handy thing to move windows around. It's the only reason I can specify exactly how large the window is. I think if you use the grids on nVidia forceware drivers, you can define specific windows sizes as well. What would be best is a macro to resize the SW window. That seems like it should be possible, but I don't know how to do it, off the top of my head. I'd have to look. For that matter, Mike's macro could be altered to control all of these issues.

Dale,

Since the model is SW2001+, shading with edges would only be the default in

2004 (since these are reversed).

While this does give some sense of the overall "system" performance, I think the waters get a bit muddied with regards to actual raw CPU performance.

Since I'll have a dual Opteron, and a single 64FX-51, all that matters is that I use the exact same thing for both.

Regards.

Mark

Awesome!!! We broke below 28!!

- Eddy

Several months ago, I asked Mike about changing document properties, and the essence of his reply was to run the tests with his default settings. I hope Mike will join in here and give us the 'official' word on this.

Art W

FYI: My initial times were with the model defaults, image and wire set to highest setting. Resolution at 1280 x 1024.

2004 SP 2.1. OS, XP Pro. I reset image and wire to lowest and got 23.7 max, 22.4 min. Lowered resolution to 1080 x 768, 22.4-22.7

Mike Eckstein