Amiga.org
Amiga computer related discussion => Amiga Hardware Issues and discussion => Topic started by: Jose on March 22, 2004, 09:46:11 PM
-
Before you think I'm mad HERE (http://www.aminet.net/docs/rview/CyberVision64.txt)is a review of the Cybervision54 card, wich has some busspeed tests at the end. It was the values of these tests that puzzled me. Wasn't ZIII limited to 10mb/s or something?
I seem to remember in the past, on aminet, a review of another ZIII card (don't remember wich, and couldn't find it) wich stated similar values in asynchrous mode and slower ones on synchronous (or was that the other way?..)
Anyway, can someone that knows clarify after all what are the speed etc. diferences on Sync. vs. async. modes and what more limitations are there for each mode?
I pretty remember that ZIII could do about 50mb/s on one of these modes 8-)
-
I have a little proggie you can test your gfx card's VRAM access speeds with, if you want to test it. Runs on 68040 or above only (unless I take the move16 tests out)...
-
Maybe you mis-read the results. The 50MB/s quoted looks like the read speed from FAST ram to me...the video card is doing:
$40000000 readw 690.1ns 2.9MByte/s
$40000000 readl 690.1ns 5.8MByte/s
$40000000 readm 693.7ns 5.8MByte/s
$40000000 writew 538.2ns 3.7MByte/s
$40000000 writel 538.7ns 7.4MByte/s
$40000000 writem 538.2ns 7.4MByte/s
Assuming $40000000 is zorro video card RAM....this would be about right for ZIII I think.
Steve
-
@JDK:
Yes, 0x40000000 is the normal base-address of the CyberVision64 card. Though if you do a test at 0x40000000 you will test the access speed for the register space of the CyberVision64, the VRAM on the other hand starts at 0x41400000 if the base address is 0x40000000.
Here are register and memory space tests for my CyberVision64 with a A3640 040 cpu-card (these test were done while the CyberVision64 was displaying the workbench):
Register space:
BusSpeedTest 0.19 (mlelstv) Buffer: 262144 Bytes, Alignment: 32768
========================================================================
memtype addr op cycle calib bandwidth
user $40000000 readw 858.1 ns normal 2.3 * 10^6 byte/s
user $40000000 readl 859.8 ns normal 4.7 * 10^6 byte/s
user $40000000 readm 853.6 ns normal 4.7 * 10^6 byte/s
user $40000000 writew 710.9 ns normal 2.8 * 10^6 byte/s
user $40000000 writel 711.2 ns normal 5.6 * 10^6 byte/s
user $40000000 writem 706.0 ns normal 5.7 * 10^6 byte/s
Memory space:
BusSpeedTest 0.19 (mlelstv) Buffer: 262144 Bytes, Alignment: 32768
========================================================================
memtype addr op cycle calib bandwidth
user $41400000 readw 961.5 ns normal 2.1 * 10^6 byte/s
user $41400000 readl 963.7 ns normal 4.2 * 10^6 byte/s
user $41400000 readm 962.1 ns normal 4.2 * 10^6 byte/s
user $41400000 writew 404.3 ns normal 4.9 * 10^6 byte/s
user $41400000 writel 404.3 ns normal 9.9 * 10^6 byte/s
user $41400000 writem 402.0 ns normal 10.0 * 10^6 byte/s
The ZorroIII bus can do a bit better with an accelerator interfacing the CPU-FastSlot better. With my CSPPC it can to do up to 15MB/Sec with the CV64 and with the original A3400 030 cpu-card I think I got close to 16MB/Sec. Needless to say the A3640 040 cpu-card does a terrible job at interfacing with just about everything :D.
/Patrik
-
@Patrik
I beginning to see the term "Busmaster" applies to people as well as hardware ;-)
Hmmm, that makes me think. I wonder where Shawn is these days? I bet he has some interesting Zorro bus benchmarks :lol:
-
Thanks for the refresher...long time since I had my CV64 :-)
Needless to say Jose was indeed looking at the wrong results for a bus speed indication.
Cheers,
Steve
-
I think the original CV64 was about the fastest GFX-card in writing to ZorroIII getting near 20 MB/s with a fast processor and following wind.
-
i seem to remember my old CV64/3D pushing between 12-25Mb sec with various tests from "bus test" on an A4000T CyberstormPPC/060...
-
I have a little proggie you can test your gfx card's VRAM access speeds with, if you want to test it.
I'm still waiting for this nice little proggie :-). Check your old PMs.
________
Gay Asian (http://www.fucktube.com/categories/610/asian/videos/1)
-
@Cass
Sorry about that - this whole framework update made me forget ;-)
Anyway, you can download it here (http://www.nyteshade.com/karlos/util/exng_util_pixeltest2004-03-23.lzx).
-edit-
I forgot to say - the program was compiled using 040+FPU options and linked with the 040+FPU version of the framework libs.
As such, it *needs* an FPU to run!
-
Thanks Karlos! :-D
________
Zoloft settlements (http://classactionsettlements.org/)
-
Yes, I was looking at the wrong results..
But it's great to know that ZIII can do much more than 10mB/s.
So does anyone know the diference between sync and async modes in performance and maye how they work and restrictions that exist in each when working with cards?
By the way, this seems to indicate poor implementation in the existing PCI to Amiga bust boards available, cause from the tests around they seem to peak at 10mb/s.
@Karlos
I don't have an A4000 to test it... but I recently got an A4000 motherboard on eBay for a very decent price;)
Cool avatar:-)Is that Beavis on the right side?
-
The theoretical limit for the Zorro III bus is
100 Mb/sec.
25Mhz * 32bit(4 Bytes) = 100Mb/sec
That is with the "HYPOTHETICAL NON EXIST BUSTER"
I remember the bla bla bla from the big C saying
they were going to fix it..
-
@Wol
What about Buster rev.11 ?
So all we need is a new rev. Buster ?!!! I can't believe nonone has done it yet... there must be something more than that that comes into play.
Who want's an A4000 with Buster rev.20 with 90mB/s ZorroIII transfer? :-D
-
Jose wrote:
Cool avatar:-)Is that Beavis on the right side?
Yes, its beavis and butthead rocking away. The original image was:
(http://www.animationlibrary.com/Animation11/Creatures_and_Cartoons/Cartoon_Characters/Beavis_6.gif)
-
Just checking, but you definately means Mb and not MB?
-
NightShade737 wrote:
Just checking, but you definately means Mb and not MB?
I'm pretty sure people here mean 2**10 bytes :-)
-
If anyone is interested in the details of the ZorroIII bus, the ZorroIII specification (http://www.thule.no/haynie/zorroiii/docs/zorro3.pdf) covers pretty much all of it.
Have fun!
/Patrik
-
I've first read about the benchmark program here (http://www.amiga.org/forums/showthread.php?t=3190). My results are the following:
640*400@24 bits
Read RAM : 47274.53 K/sec
Write RAM : 46382.56 K/sec
RAM->RAM : 22630.56 K/sec
RAM->RAM[16] : 26444.66 K/sec
Read VRAM : 3220.04 K/sec
Write VRAM : 7284.17 K/sec
RAM->VRAM : 7065.75 K/sec
RAM->VRAM[16] : 6811.73 K/sec
VRAM->RAM : 2841.36 K/sec
VRAM->RAM[16] : 3157.89 K/sec
Conversion : 21386.14 K/sec [output bandwidth]
Conversion attained 302.67% copy speed
1024*768@16 bits
Read RAM : 47571.85 K/sec
Write RAM : 47164.18 K/sec
RAM->RAM : 22111.55 K/sec
RAM->RAM[16] : 26444.66 K/sec
Read VRAM : 2985.07 K/sec
Write VRAM : 7330.83 K/sec
RAM->VRAM : 7024.39 K/sec
RAM->VRAM[16] : 6923.08 K/sec
VRAM->RAM : 2772.64 K/sec
VRAM->RAM[16] : 3027.75 K/sec
Conversion : 7072.69 K/sec [output bandwidth]
Conversion attained 100.69% copy speed
All this started from a real problem (http://amigaworld.net/modules/newbb/viewtopic.php?topic_id=3056&forum=9&start=20&viewmode=flat&order=0): I run Earth 2140 and noted jerky scrolling. I thought that is impossible to have in market s/w that can't be used on the h/w it's supposed to work on.
If I calculated correctly, for a 16bp 800x600 @ 30fps you'll need about 30MB/sec!
________
Vapolution (http://www.vaporshop.com/vapolution-vaporizer.html)
-
Please ignore the conversion speed results for this test - there's clearly a bug in the calculation for 24-bit pixels (forgot to divide by bytes per pixel there, hence 3x expected). Besides, only the read / write / copy to and from VRAM give the actual bus performance estimates.
@Cass,
Well, that figure you calculate for 800x600x16 assumes that you are basically copying every frame from fast ram to video ram, which I strongly doubt. Only movie players would need to do this.
I would expect any well written 2D tile based games to cache background stuff as bitmaps (where transparency is not an issue) and only refresh the areas of the display where stuff happens.
Even scrolling (without using a superbitmap approach) should only amount to a blit of the remaining visible background area and then an update of all the smaller areas that have changed / been uncovered. I've experimented with this basic idea on the bvision and can confirm that its fast, even on 040.
-edit-
Back on the bus speed issue, here are some results for the local bus on a BlizzPPC 060 @ 50MHz + BVisionPPC:
[size=x-small]
Surface width: 640, height: 480, modulus: 0
Surface hwWidth: 640, hwHeight: 480
Test data pixel format
Bytes : 2, endian native
Bits : A[ 0] R[ 5] G[ 6] B[ 5]
Offsets : A[ 0] R[ 11] G[ 5] B[ 0]
Maxima : A[ 0] R[ 31] G[ 63] B[ 31]
Window pixel format
Bytes : 2, endian native
Bits : A[ 0] R[ 5] G[ 6] B[ 5]
Offsets : A[ 0] R[ 11] G[ 5] B[ 0]
Maxima : A[ 0] R[ 31] G[ 63] B[ 31]
Read RAM : 38652.13 K/sec
Write RAM : 37944.66 K/sec
RAM->RAM : 16699.80 K/sec
RAM->RAM[16] : 21980.19 K/sec
Read VRAM : 5708.84 K/sec
Write VRAM : 17526.77 K/sec
RAM->VRAM : 12143.51 K/sec
RAM->VRAM[16] : 11964.10 K/sec
VRAM->RAM : 4606.52 K/sec
Conversion : 12121.21 K/sec [output bandwidth]
Conversion attained 99.81% copy speed
[/size]
-
@Karlos
Actually, that's what I thought in the first place, but I had my doubts after the poster's suggestion and the facts (the game itself)... :-?
Anyhow, the 7MB/sec it's pityful and I don't know if it's idiopathic, or there something that can be done to improve it (I've already have a BUSter v11).
________
Pissing fetish (http://www.fucktube.com/categories/822/fetish/videos/1)
-
@Cass:
7MB/s? It sounds as if you were using a CV64/3D... real CV64s are faster than that (CV64/3D is a cheap design done quickly, just compare the number of components)... I got more than 9MB/s with my Picasso4 and it has a slow z3 bus...
It's a pity that no one has released an updated (and fixed) buster... it would make normal Z3 cards and PCI solutions much faster...
-
i also seem to remember is that the A4000 is based on the A3000 design which in turn is based on supporting 030 chips and bus bursts etc. to from zorro space.
the reason you get good bench results with an 030, is that the bus interface was designed for an 030.
the commodore 040 card didn't support bursting modes, or various other things, so the worst CPU card to bench with is the A3640. the main changes in the rev11 buster over the rev9 were to support the crappy bus signal from the A3640 card.
phase 5 built their cards based on this design and had some interface logic between the 040/060/PPC chip and 030 bus to properly support the various burst modes etc.
from what i remember, this was the major update between the cyberstorm1 and cyberstorm2 series logic, and continued into the cyberstorm3/PPC range.
not sure about any other accelerator manufacturers, but this can easily be found in the speed the CPU can access chip ram.
with my EC030, cybII040, and cybIII060/PPC it was around 7Mb/s, but with my Warpengine040 (28Mhz overclocked to 40Mhz! yehaa! big fat heatsink & fan, attached with heat xfer compound and superglue - FPU thought it didn't exist tho it still worked(?) ;) i was looking at about 2.5Mb/s
(buster rev9 and rev11)
so from my expirience, the best cards to bench with are either 030's or phase5 designed cards on A4000's
-
Do you think it would be expensive to get someone to do a new proper buster chip? Or anyone left that could do it, for that matter.. :-(
-
Buster rev.12... That would be cool... :)