How fast is the Fastest 68882?

[derringer3]

Recently i decided that i will speed up my mtec 68030 board. It is the version for a500/1000/2000. Originally it contains a 68030@14,2Mhz and a 68882@20MHz. (the chips are both clocked at 16MHz by factory.)

Unfortunatelly only the FPU speed can be changed. SO bad...
So the result is:
68882 at 20Mhz: 0.48Mflops
68882 at 24Mhz: 0.51Mflops
68882 at 27Mhz: 0.58Mflops
68882 at 32Mhz: 0.68Mflops
68882 at 40Mhz: 0.79Mflops

Sysinfo v3.24 and Sysspeed says exactly the same numbers in Mflops, but in MIPs they say that my 68030 is 2.61Mips and 4,48Mips. Strange.

AIBB show the difference but it say that the difference between 68882@20Mhz and 68882@40Mhz is only 2-3%. Anyway AIBB shows the correct MHz datas for 20,24,27,32 settings, but when i use 40Mhz oscillator it says that is only clocked at 33,4MHz. I think it is wrong because the difference in Mflops reveals the real difference.

In real life test, for example gloom deluxe, is running much faster at 40Mhz than 20Mhz.  


So my question is what was the fastest 68882 ever made, and how many Flops it can produce?

[fiat1100d]

Factory fastest 68882 (not overclocked) is 50 MHz. It gives the following results on my Blizzard 1230-IV:

- SysInfo: 1.33 Mflops (68030@50 9.67 Mips - reported as 51.80 MHz)
- AIBB: 0.8978 Scalar AIBB MFlops (Flops test, CP Math). Correctly reports 50 MHz clock

[Piru]

The 68882 speed scales linear to the FPU clock.

"Flops" is meaningless in measuring FPU speed anyway. Use SPECfp.

[kidkoala]

@all

I have a few questions about cpu's on accelerators as well;
I have a cd32 with sx32mk2 (think it's a 40Mhz cpu), and I bought a fpu@33mhz, and i bought a crystal @33mhz as well.
Now, can you people explain to me;

1. what is the connection between the fpu and the crystal? (something about synchronous/asynchronous?), and can i just put in the fastest fpu and crystal there is? (says something about max. 33mhz in my primitive papers on the sx32).

2. can i put in the fastest cpu i find? (like a 50mhz or something), is there a connection between the cpu and the fpu speed?

3. what commands or programs do i use to find out the speeds that are right now? and how do i overclock the cpu/fpu?

4. what is the fpu used for (i know about floating-point operations, but what is that, and give me some examples of which programs/games use it and which doesn't)?

hope someone wants to answer all this, please give thorough answers if you do:) thanks a lot!

[alexh]

Piru wrote:
"Flops" is meaningless in measuring FPU speed anyway. Use SPECfp.

Not in this case. YES if you are comparing two totally different FP units with different architectures. The SAME FP, with the SAME benchmark tool, just different clock speeds, FlOPS will be an accurate benchmark.

[Piru]

You could aswell just use the MHz value directly.

Anyway, "flops" is no real benchmark, it typically benchmarks fadd in loop or so, hardly gives any real indication of the FPU performance (which SPECfp does).

[countzero]

honestly I don't think upgrading an fpu will make your amiga make any faster. As far as I know, no games make use of it, and you'll find it useful only if you're using 3d graphic applications or databases programs.

[Karlos]

derringer3 wrote:
Recently i decided that i will speed up my mtec 68030 board. It is the version for a500/1000/2000. Originally it contains a 68030@14,2Mhz and a 68882@20MHz. (the chips are both clocked at 16MHz by factory.)

Unfortunatelly only the FPU speed can be changed. SO bad...
So the result is:
68882 at 20Mhz: 0.48Mflops
68882 at 24Mhz: 0.51Mflops
68882 at 27Mhz: 0.58Mflops
68882 at 32Mhz: 0.68Mflops
68882 at 40Mhz: 0.79Mflops

Sysinfo v3.24 and Sysspeed says exactly the same numbers in Mflops, but in MIPs they say that my 68030 is 2.61Mips and 4,48Mips. Strange.

AIBB show the difference but it say that the difference between 68882@20Mhz and 68882@40Mhz is only 2-3%. Anyway AIBB shows the correct MHz datas for 20,24,27,32 settings, but when i use 40Mhz oscillator it says that is only clocked at 33,4MHz. I think it is wrong because the difference in Mflops reveals the real difference.

In real life test, for example gloom deluxe, is running much faster at 40Mhz than 20Mhz.  

This is bizzare. I am sure gloom deluxe made no use of the FPU.

Is it possible that your CPU clock is derived from the same crystal?

[derringer3]

I think for gloom deluxe it is not required an fpu. But i think it uses if it found. Whoever It is much faster when i doubled the fpu clock. At the same speed i can run it on twice larger window, with 1*1 pixel. And it is only the demo version.

Anyway in winxp there is a program called fraps, it can show any srceen's FPS. Its quiet good application to measure some 3d games. Any similar amiga program?

Unfortunatelly the Cpu don't use the crystal on the turbo board. It is uses the a500 clock i think. Just the Fpu use it. The Cpu clock constantly 14.2Mhz no matter which crystal i used. (If somebody do a hack for faster cpu let me now.)

[alexh]

Piru wrote:
You could aswell just use the MHz value directly.

I dont think you can. Adding a faster FPU may make the FPU finish processing it's instruction VERY fast, but end up sitting there waiting for the next instruction / data. Its effectiveness also depends on things like 68k bus speed etc.

[derringer3]

Have anybody ever try it on 66MHz?

Or anybody can tell me which the higher -stable- Mhz of an 68882/50Mhz?