| 2 MB /sec Network Speed | page 1 2
|
|---|
|
|---|
| | 
Simo Koivukoski
(Apollo Team Member)
Posts 601
26 Sep 2016 15:46
| gregthe canuck wrote:
|
Would be interesting to see if Roadshow 68K was any faster.
|
A4kt + Roadshow: EXTERNAL LINK
| |
| | 
Marcus Gerards
Posts 58
26 Sep 2016 19:16
| Simo Koivukoski wrote:
|
gregthe canuck wrote:
|
Would be interesting to see if Roadshow 68K was any faster.
|
A4kt + Roadshow: EXTERNAL LINK
|
What's the point of posting these values? You know that TCP/IP-performance on the Amiga depends on the CPU speed. And what CPU card did the machine behind that link have? :)
| |
| |
Marcus Gerards
Posts 58
26 Sep 2016 19:17
| Szyk Cech wrote:
|
Maybe that will be better to concentrate every further efforts for stand alone...
|
Let me think about it for a moment. Erhm...no.
| |
| |
Simo Koivukoski
(Apollo Team Member)
Posts 601
26 Sep 2016 19:47
| Marcus Gerards wrote:
|
Simo Koivukoski wrote:
|
gregthe canuck wrote:
|
Would be interesting to see if Roadshow 68K was any faster.
|
A4kt + Roadshow: EXTERNAL LINK
|
What's the point of posting these values? You know that TCP/IP-performance on the Amiga depends on the CPU speed. And what CPU card did the machine behind that link have? :)
|
No idea of cpu, but it has Z3 bus. :)
| |
| |
Simo Koivukoski
(Apollo Team Member)
Posts 601
27 Sep 2016 17:34
| A comparison between the x10 and x14 cores with X-Surf-100 Zorro-II:V500_3499_x10 (PAL 7.093795 MHz x 10 = 70.9MHz):
Packet size 1k bytes: 1662.61 KByte/s Tx, 1813.20 KByte/s Rx.
Packet size 2k bytes: 1805.81 KByte/s Tx, 2000.95 KByte/s Rx.
Packet size 4k bytes: 1936.80 KByte/s Tx, 2081.89 KByte/s Rx.
Packet size 8k bytes: 1970.21 KByte/s Tx, 2006.31 KByte/s Rx.
Packet size 16k bytes: 1993.55 KByte/s Tx, 2081.89 KByte/s Rx.
Packet size 32k bytes: 1988.70 KByte/s Tx, 2150.89 KByte/s Rx.V500_3478_x14_895 (PAL 7.093795 MHz x 14 = 99.3MHz):
Packet size 1k bytes: 1792.51 KByte/s Tx, 2013.84 KByte/s Rx.
Packet size 2k bytes: 1925.54 KByte/s Tx, 2146.03 KByte/s Rx.
Packet size 4k bytes: 2072.29 KByte/s Tx, 2199.50 KByte/s Rx.
Packet size 8k bytes: 2131.46 KByte/s Tx, 2232.50 KByte/s Rx.
Packet size 16k bytes: 2148.60 KByte/s Tx, 2247.70 KByte/s Rx.
Packet size 32k bytes: 2111.17 KByte/s Tx, 2243.38 KByte/s Rx.  AmigaOS 3.1 / AmiTCP-3.0b2 / NETIO Benchmark
| |
| | 
Szyk Cech
Posts 191
28 Sep 2016 14:36
| How is this possible for Amiga to be so fast?!?
| |
| | 
Cunn Pole
Posts 29
28 Sep 2016 16:22
| Is it just me or is that case exceptionally clean?
| |
| | 
Mr-Z EdgeOfPanic
Posts 189
29 Sep 2016 05:29
| Looking at the benchmarks there is not much difference between a x10 and x14 clocked vampire.
This means that the Vampire is pretty much maxing out the Zorro II bus.
| |
| | 
Thierry Atheist
Posts 644
29 Sep 2016 06:44
| Yes... Seems that even if the Vampire 2 Apollo was operating at 600 MHz, the AMIGA motherboard can't operate the port any faster. It's nothing to be concerned about, as the standalone Vampire WILL ROCK!!! It should EASILY be able to get up to 48 megabits (6 Megabytes) a second transfer rates!
| |
| | 
Amiga 4Life
Posts 101
29 Sep 2016 11:22
| yeah, that looks good right there, Apollo seems to bring out the best in Zii boards. I would like to know what
precent of the cpu (apollo) is being used to drive X-Surf-100 to that speed ?..The other cpu isn't able to push
as much data down the bus, something is not setup correctly, and thats a a4kT ...(thats just me)..
| |
| |
Simo Koivukoski
(Apollo Team Member)
Posts 601
29 Sep 2016 13:55
| CPU usage for x10 core looks like this. CPU usage decreases when receiving larger packet size. Receiving 32k packet uses about 50% CPU. Sending side uses always about 95% CPU. 
| |
| | 
John Heritage
Posts 111
01 Oct 2016 03:19
| Gunnar - does the A1200 PCMCIA implementation (And IDE implementation) have the same limitations as the A600?
| |
| | 
Gregthe Canuck
Posts 274
01 Oct 2016 20:44
| Hi Simo - thanks for doing those benchmarks. It would be interesting for comparison purposes to see how Roadshow performs under similar conditions. I am curious to see if Roadshow really is more or less efficient than Amitcp. Cheers!
| |
|
Documentation about the Vampire hardware