2D Graphics Performance

[zakee00]

um...
test 1: 5.7seconds
test 2: 5.7seconds
1407MPixles/sec
5628MB/sec memory bandwidth. ill try OCing my video card and see if that raises the scores.
right now im default Ultra speeds.

EDIT: there is something wrong with the program, my scores are totally bogus. i ran it again, and i got 72.2secs. then i ran test 2 and it said 72.2 secs also, when it was really much shorter time then test 1...

 

[zephyrprime]

Originally posted by: xtknight
If I'm right that GDI isn't accelerated and you want extremely high-performance, you may want to use OpenGL instead.

GDI is accelerated.

 

[Sparky19692]

specs are below
Test 1 104 sec
Test 2 1588 mp/sec 6352 meg/sec

 

[RonS]

Wowl Zebo has the fastest times so far, by a pretty huge margin. The 6800 cards are the fastest cards as a group. Zebo's rotate test is about 2 times faster than the average, and the BitBlt test is about 3 times faster than the average.

Zakee00 - yes, there is a bug, sort of. I grab the Device Context (DC) for the window just once, and then rotate the bitmap (PLGBLT) multiple times. If the program looses focus, it can cause the DC to become invalid and the test terminates early. I've noticed that the DC can become invalid even when not moving the focus -- since this is just a quick-n-dirty program, I didn't bother digging to figure out why.

302efi - I'm assuming, based on your scores, that you're also running a 6800. Right?

 

[302efi]

Ron S - Yes I run a 6800 Ultra extreme.

System Specs:

A64 3000+ 754 (512 kb)
Soltek NF3 250gb
2x256 Geil Ultra value
EVGA 6800 Ultra Extreme


*** All the parts are run at there stock speeds

 

[dug777]

heh ran it my folks dell 8200

with p4 2ghz (nwood), gf4mx420:

1st test: 164.2
2nd test: 225/900

(but at 1024 not 1280)

 

[RonS]

at 1280x1024, the first test would be even worse.

 

[dug777]

hardly ne diff in fact, greater change in the 2nd

@1280

1st test: 164.4
2nd test: 222/888

My 9800 pro (AXP 2000+ @2600+, 512MB, A7V8X) rig gave

1st test: 125.8
2nd test: 1296/5184

 

[RonS]

What I meant was the the first test draws outside the window bounds if the resolution is less than 1280x1024. So at lower resolutions, fewer pixels are being drawn so the test will finish faster.

With the second test, it doesn't really matter what the resolution is set to. It's simply a measure of video memory bandwith. If you reduce the number if bits/pixel, you should see the MegaPixels/sec value increase but the MegaBytes/sec value should stay the same.

 

[Squally Leonharty]

Erm, just wondering, but why have I not received a PM yet? *_*

 

[TStep]

P4 2.4/533
Abit TH7II
1Gig PC800
Geforce 3 (core 200 / memory 460(230 DDR), 4X AGP, 64meg DDR)
Monitor's max resolution was only 1024x768

Test 1: Rotate Bitmap - 131.2 sec
Test 2: 484 MPixels/sec, 1936 MB/sec memory bandwidth

 

[RonS]

Squally - I just re-checked. As far as I can tell, I've responded to all of the PM's I've received. Can you PM me again?

 

[Todd33]

OP should edit the OP with ongoing results, it's a waste of time to scroll through the thread and find nothing.

 

[RonS]

Good idea - I'll do this tonight with whatever results I get back today, and then call my little experiement done.

 

[Squally Leonharty]

Originally posted by: RonS
Squally - I just re-checked. As far as I can tell, I've responded to all of the PM's I've received. Can you PM me again?

Should be fine now, I hope. >_o

 

[Squally Leonharty]

Sorry for double-posting. The results at 1600x1200 are as follows:

Test 1: 86.5 secs
Test 2: 855 lines/sec, 1479 MPixels/sec, 5916 MB/sec memory bandwidth

I'm not exactly sure if the second test is supposed to measure the average or the max, because I've seen it going higher than that, often at 1500 MPixels/sec, 6000 MB/sec memory bandwidth.

PC specs:
AMD Athlon64 3500+ @ 2.2GHz
Asus A8V Deluxe rev. 2.00
Crucial Ballistix PC3200 @ 2-2-2-8-1T
Club-3D Radeon X800 XT (flashed&clocked at XT PE speeds: 520/560)

 

[Zebo]

Originally posted by: RonS
Wowl Zebo has the fastest times so far, by a pretty huge margin. The 6800 cards are the fastest cards as a group. Zebo's rotate test is about 2 times faster than the average, and the BitBlt test is about 3 times faster than the average.

Zakee00 - yes, there is a bug, sort of. I grab the Device Context (DC) for the window just once, and then rotate the bitmap (PLGBLT) multiple times. If the program looses focus, it can cause the DC to become invalid and the test terminates early. I've noticed that the DC can become invalid even when not moving the focus -- since this is just a quick-n-dirty program, I didn't bother digging to figure out why.

302efi - I'm assuming, based on your scores, that you're also running a 6800. Right?

I'm wildly overlcoked ron with LL mem. I'm not really understanding this test though and how it relates to programs. but I played for another data point for ya.

If you want I can host program instead of emailing dozens...

 

[Zebo]

Here's another

Athlon XP 3200 with GF3-Ti200


Rotate bitmap = 105.5 sec

ScrollTest saw: 439 MPixels/sec... 1760 MB/sec memory bandwidth.

 

[roguerower]

Just ran it on my ATI X800 Pro no OC
First test time: 19.3 secs
Rotate took: 19.3
Scroll test saw 1018 Mpixels/sec, 4072 MB/sec Memory bandwith

 

[Zebo]

Originally posted by: roguerower
Just ran it on my ATI X800 Pro no OC
First test time: 19.3 secs
Rotate took: 19.3
Scroll test saw 1018 Mpixels/sec, 4072 MB/sec Memory bandwith

impossible, you guys arnt letting test run alone. if you switch to another proggy time resets.

 

[Chaotic42]

All tests with Radeon 9700 non-pro

Speed: Stock
Window: 1600x1200
RotateBitmap-146.5s
Scrolltest-874 MPixels/s, 3496 MB/s memory bandwidth

Speed: 371.25/563
Window: 1600x1200
RotateBitmap-147.7s
Scrolltest-874 MPixels/s, 3496 MB/s memory bandwidth

Speed: Stock
Window: 1920x1440
RotateBitmap-147.6s
Scrolltest-914 MPixels/s, 3656 MB/s memory bandwidth

Speed: Stock
Window: 3200x1440 (spanned over two monitors, one 1920x1440, one 1280x960)
RotateBitmap-148.6s
Scrolltest-994 MPixels/s, 3976 MB/s memory bandwidth


Speed: Stock
Window: 3328x1536 (spanned over two monitors, one 2048x1536, one 1280x960)
RotateBitmap-147.8s
Scrolltest-974 MPixels/s, 3896 MB/s memory bandwidth

 

[RonS]

Chaotic42 - your results confirm what I said earlier that resolution past 1280x1024 won't change the results very much. Same RotateBitmap and memory bandwidth numbers.

 

[Elfear]

Here is what I got:

1st Test- 74sec
2nd Test- 1393 MPixel/sec & 5572 MB/sec

 

[dug777]

jeez, i know chaotic's 9700 is a non-pro, but the difference b/w my scores w/9800 pro and his kard are massive...sure there isn't something else @play here

I had noticed that even on the second test my cpu usuage went up 2 100%, so i suspect that there's a limiting effect there- seems to be a logical assumption yes/no?

 

[xtknight]

Originally posted by: zephyrprime

Originally posted by: xtknight
If I'm right that GDI isn't accelerated and you want extremely high-performance, you may want to use OpenGL instead.

GDI is accelerated.

It seems as though my tests are confirming otherwise.
P4 with HyperThreading (50% usage=100% on Athlon)

Surely this program shouldn't be using up all of my processor if the video card were taking care of the processing. If that were so, games would run like !@#@ :Q I wish there were a program to look at GPU load, then we'd know for sure. Maybe what you're saying is GDI acceleration requires extra code that this program didn't have in it.