TT and NemBench

[agranlund]

Here's my Atari 520STM.
Not quite a TT, but it likes to pretend that it is and who am I to argue

It's equipped with a TF534 accelerator running @ 50Mhz.
4MB TT-RAM on accelerator card.
4MB ST-RAM on mainboard.

I don't have any FPU that can handle 50Mhz so the FPU slot is empty until I find one.

Code: Select all

NemBench v2.1 - precision CPU/FPU profiler.

Integer multiply (16bit)     -> 1.921 Mips (~313%)
Integer divide (16bit)       -> 1.135 Mips (~313%)
Linear (stalled) integer     -> 24.975 Mips (~313%)
Interleaved (piped) integer  -> 24.975 Mips (~313%)

16bit read (100% hit)        -> 24.630 MByte/sec (~313%)
16bit write (100% hit)       -> 14.492 MByte/sec (~240%)
32bit read (100% hit)        -> 49.261 MByte/sec (~313%)
32bit write (100% hit)       -> 28.901 MByte/sec (~433%)

Linear 32bit read (ST-Ram)   -> 3.919 MByte/sec (~73%)
Linear 32bit write (ST-Ram)  -> 3.921 MByte/sec (~60%)
Linear 32bit copy (ST-Ram)   -> 1.964 MByte/sec (~60%)

Linear 32bit read (FastRAM)  -> 30.131 MByte/sec (~566%)
Linear 32bit write (FastRAM) -> 27.976 MByte/sec (~433%)
Linear 32bit copy (FastRAM)  -> 13.988 MByte/sec (~433%)

[Cyprian]

really impressive figures, just for the record, it should be posted there Falcon and NemBench.
Here we compare vs TT and Nembech shows figures vs Falcon.

[soviet9922]

do you have a picture of how you installed the tf534 inside the case ?

[frank.lukas]

Here's my Atari 520STM.
Not quite a TT, but it likes to pretend that it is and who am I to argue :)

It's equipped with a TF534 accelerator running @ 50Mhz.
4MB TT-RAM on accelerator card.
4MB ST-RAM on mainboard.

I don't have any FPU that can handle 50Mhz so the FPU slot is empty until I find one.

Code: Select all

NemBench v2.1 - precision CPU/FPU profiler.

Integer multiply (16bit)     -> 1.921 Mips (~313%)
Integer divide (16bit)       -> 1.135 Mips (~313%)
Linear (stalled) integer     -> 24.975 Mips (~313%)
Interleaved (piped) integer  -> 24.975 Mips (~313%)

16bit read (100% hit)        -> 24.630 MByte/sec (~313%)
16bit write (100% hit)       -> 14.492 MByte/sec (~240%)
32bit read (100% hit)        -> 49.261 MByte/sec (~313%)
32bit write (100% hit)       -> 28.901 MByte/sec (~433%)

Linear 32bit read (ST-Ram)   -> 3.919 MByte/sec (~73%)
Linear 32bit write (ST-Ram)  -> 3.921 MByte/sec (~60%)
Linear 32bit copy (ST-Ram)   -> 1.964 MByte/sec (~60%)

Linear 32bit read (FastRAM)  -> 30.131 MByte/sec (~566%)
Linear 32bit write (FastRAM) -> 27.976 MByte/sec (~433%)
Linear 32bit copy (FastRAM)  -> 13.988 MByte/sec (~433%)

... roughly the same values with my PAK68/3-030 50Mhz and FRAK/2

Code: Select all

NemBench v2.1 - precision CPU/FPU profiler.

Integer multiply (16bit)     -> 1.921 Mips (~313%)
Integer divide (16bit)       -> 1.135 Mips (~313%)
Linear (stalled) integer     -> 24.975 Mips (~313%)
Interleaved (piped) integer  -> 24.975 Mips (~313%)

Float multiply (64bit)       -> 0.839 MegaFlops (~316%)
Float divide (64bit)         -> 0.544 MegaFlops (~314%)
Linear (stalled) float       -> 1.685 MegaFlops (~316%)
Interleaved (piped) float    -> 1.665 MegaFlops (~312%)

16bit read (100% hit)        -> 24.630 MByte/sec (~313%)
16bit write (100% hit)       -> 14.285 MByte/sec (~237%)
32bit read (100% hit)        -> 49.261 MByte/sec (~313%)
32bit write (100% hit)       -> 28.571 MByte/sec (~428%)

Linear 32bit read (ST-Ram)   -> 3.917 MByte/sec (~73%)
Linear 32bit write (ST-Ram)  -> 3.917 MByte/sec (~60%)
Linear 32bit copy (ST-Ram)   -> 1.682 MByte/sec (~52%)

Linear 32bit read (FastRAM)  -> 31.282 MByte/sec (~588%)
Linear 32bit write (FastRAM) -> 27.507 MByte/sec (~426%)
Linear 32bit copy (FastRAM)  -> 9.630 MByte/sec (~298%)

[LaoMu]

Hi Tuxie,

great results. By the way I am still waiting on the VoFa.

BR lao Mu

[Rustynutt]

Need to dig out the old Mighty Sonic 32 @44mhz and NOVA ATI Nembench files. Sorry Amiga

[czietz]

And since I'm benchmarking, here's my TT accelerated to 48 MHz. Much like with Tuxie's result from 2019, no CaTTamaran is involved.

Code: Select all

NemBench v2.1 - precision CPU/FPU profiler.

Integer multiply (16bit)     -> 1.855 Mips (~302%)
Integer divide (16bit)       -> 1.096 Mips (~302%)
Linear (stalled) integer     -> 24.165 Mips (~303%)
Interleaved (piped) integer  -> 24.165 Mips (~303%)

Float multiply (64bit)       -> 0.811 MegaFlops (~306%)
Float divide (64bit)         -> 0.526 MegaFlops (~304%)
Linear (stalled) float       -> 1.338 MegaFlops (~251%)
Interleaved (piped) float    -> 1.336 MegaFlops (~251%)

16bit read (100% hit)        -> 23.809 MByte/sec (~303%)
16bit write (100% hit)       -> 16.129 MByte/sec (~268%)
32bit read (100% hit)        -> 47.619 MByte/sec (~303%)
32bit write (100% hit)       -> 32.258 MByte/sec (~483%)

Linear 32bit read (ST-Ram)   -> 7.972 MByte/sec (~150%)
Linear 32bit write (ST-Ram)  -> 7.848 MByte/sec (~121%)
Linear 32bit copy (ST-Ram)   -> 3.938 MByte/sec (~122%)

Linear 32bit read (FastRAM)  -> 26.109 MByte/sec (~491%)
Linear 32bit write (FastRAM) -> 31.170 MByte/sec (~483%)
Linear 32bit copy (FastRAM)  -> 8.131 MByte/sec (~251%)

[stormy]

And since I'm benchmarking, here's my TT accelerated to 48 MHz. Much like with Tuxie's result from 2019, no CaTTamaran is involved.

Code: Select all

NemBench v2.1 - precision CPU/FPU profiler.

Integer multiply (16bit)     -> 1.855 Mips (~302%)
Integer divide (16bit)       -> 1.096 Mips (~302%)
Linear (stalled) integer     -> 24.165 Mips (~303%)
Interleaved (piped) integer  -> 24.165 Mips (~303%)

Float multiply (64bit)       -> 0.811 MegaFlops (~306%)
Float divide (64bit)         -> 0.526 MegaFlops (~304%)
Linear (stalled) float       -> 1.338 MegaFlops (~251%)
Interleaved (piped) float    -> 1.336 MegaFlops (~251%)

16bit read (100% hit)        -> 23.809 MByte/sec (~303%)
16bit write (100% hit)       -> 16.129 MByte/sec (~268%)
32bit read (100% hit)        -> 47.619 MByte/sec (~303%)
32bit write (100% hit)       -> 32.258 MByte/sec (~483%)

Linear 32bit read (ST-Ram)   -> 7.972 MByte/sec (~150%)
Linear 32bit write (ST-Ram)  -> 7.848 MByte/sec (~121%)
Linear 32bit copy (ST-Ram)   -> 3.938 MByte/sec (~122%)

Linear 32bit read (FastRAM)  -> 26.109 MByte/sec (~491%)
Linear 32bit write (FastRAM) -> 31.170 MByte/sec (~483%)
Linear 32bit copy (FastRAM)  -> 8.131 MByte/sec (~251%)

Hi czietz,

By the way thank you so much for your lightning VME drivers they're absolutely excellent. Just wondering, do you worry about your TT being overclocked, or are the risks negligible. Thx

[czietz]

do you worry about your TT being overclocked, or are the risks negligible.

Overclocking is limited to CPU, FPU and TT-RAM. CPU and FPU run cool and stable at 48 MHz (extensively tested) and the Storm TT has been optimized to handle that speed. Therefore: no, I don't worry.