home *** CD-ROM | disk | FTP | other *** search
- Path: everest.vol.it!news
- From: bizzetti@mbox.vol.it (Fabio Bizzetti)
- Newsgroups: comp.sys.amiga.advocacy,comp.sys.amiga.hardware,comp.sys.amiga.misc,comp.sys.amiga.games,comp.sys.amiga.graphics
- Subject: Re: Why are europeans dumb enough to buy amigas?
- Date: 14 Mar 1996 02:40:40 GMT
- Organization: Video On Line
- Distribution: world
- Message-ID: <4988.6647T93T2611@mbox.vol.it>
- References: <jsheehyDo2FA3.FrH@netcom.com>
- NNTP-Posting-Host: molcl10.vol.it
- X-Newsreader: THOR 2.22 (Amiga;TCP/IP)
-
-
- >Maybe iCOMP is a CPU benchmark, and Dhrystone is a CPU/Memory benchmark.
- >A DX33 and a DX2/66 have the same speed outside the CPU, so only the cpu
- >section benefits on the DX2. Look at the 68040 on the A3000 and the
- >A4000. The 040 is a faster CPU than the 030, but the 030 can write to the
- >chipram and zorro buses faster. You have to differentiate between CPU
- >benchmarks, and system benchmarks.
-
- >John Sheehy <jsheehy@netcom.com>
-
- The 68040 isn't "worse" than 68030 in memory access times, the problem is the
- A4000/040 standard CPU board.
-
- Some memory test results:
-
- A4000/040 25Mhz (standard CPU board)
-
-
- BusSpeedTest 0.07 (mlelstv) Buffer: 16384 Bytes
- ==================================================
- loop overhead: 5.4ns
- register move: 39.0ns
- memtype op cycle bandwidth
- fast readw 155.2ns 12.9MByte/s
- fast readl 307.9ns 13.0MByte/s
- fast readm 304.7ns 13.1MByte/s
- fast writew 284.6ns 7.0MByte/s
- fast writel 572.2ns 7.0MByte/s
- fast writem 569.5ns 7.0MByte/s
- chip readw 1053.2ns 1.9MByte/s
- chip readl 1054.0ns 3.8MByte/s
- chip readm 1054.0ns 3.8MByte/s
- chip writew 1052.9ns 1.9MByte/s
- chip writel 1052.3ns 3.8MByte/s
- chip writem 1052.2ns 3.8MByte/s
-
-
- Thanks to Grypas's kind permission ;-)
-
-
- and, Blizzard1260 50Mhz, 60ns Ram:
-
- BusSpeedTest 0.07 (mlelstv) Buffer: 16384 Bytes
- ==================================================
- loop overhead: 0.8ns
- register move: 15.5ns
- memtype op cycle bandwidth
- fast readw 51.4ns 38.9MByte/s
- fast readl 79.8ns 50.1MByte/s
- fast readm 83.3ns 48.0MByte/s
- fast writew 64.7ns 30.9MByte/s
- fast writel 114.3ns 35.0MByte/s
- fast writem 123.1ns 32.5MByte/s
- chip readw 899.5ns 2.2MByte/s
- chip readl 899.4ns 4.4MByte/s
- chip readm 899.6ns 4.4MByte/s
- chip writew 899.4ns 2.2MByte/s
- chip writel 899.5ns 4.4MByte/s
- chip writem 899.6ns 4.4MByte/s
- rom readw 52.0ns 38.5MByte/s
- rom readl 78.5ns 50.9MByte/s
- rom readm 83.8ns 47.7MByte/s
-
-
- now a Blizzard1230-IV :)
-
- BusSpeedTest 0.07 (mlelstv) Buffer: 16384 Bytes
- ==================================================
- loop overhead: 4.6ns
- register move: 40.8ns
- memtype op cycle bandwidth
- fast readw 163.8ns 12.2MByte/s
- fast readl 164.1ns 24.4MByte/s
- fast readm 151.2ns 26.4MByte/s
- fast writew 120.5ns 16.6MByte/s
- fast writel 121.2ns 33.0MByte/s
- fast writem 125.8ns 31.8MByte/s
- chip readw 892.3ns 2.2MByte/s
- chip readl 892.2ns 4.5MByte/s
- chip readm 633.2ns 6.3MByte/s
- chip writew 571.1ns 3.5MByte/s
- chip writel 570.9ns 7.0MByte/s
- chip writem 571.1ns 7.0MByte/s
-
-
- and now a TQM-1250!!! =) (a 68030 50Mhz board diffused in Italy):
-
- BusSpeedTest 0.07 (mlelstv) Buffer: 16384 Bytes
- ==================================================
- loop overhead: 4.6ns
- register move: 41.1ns
- memtype op cycle bandwidth
- fast readw 187.5ns 10.7MByte/s
- fast readl 187.4ns 21.3MByte/s
- fast readm 157.0ns 25.5MByte/s
- fast writew 143.3ns 14.0MByte/s
- fast writel 143.7ns 27.8MByte/s
- fast writem 131.4ns 30.4MByte/s
- chip readw 897.8ns 2.2MByte/s
- chip readl 898.4ns 4.5MByte/s
- chip readm 740.1ns 5.4MByte/s
- chip writew 898.0ns 2.2MByte/s
- chip writel 897.7ns 4.5MByte/s
- chip writem 716.7ns 5.6MByte/s
-
-
- well, I ran out of boards ;)
-
- my conclusion is that if all boards had the ChipMem access of Blizzard1230-IV,
- then things would be better. The A4000/040 standard CPU board is *SLOW* :-(
- I makes the 68040 a wasted ╡processor.
-
- Please note that all tests, besides A4000 one, have been made using a PAL
- screen, 1 bitplane. The A4000 test has been made with Multiscan mode, I dont
- know which color deep.
-
-
- HejDσ
-
-
- ------------------------------------------------------------------------------
- | |
- | Stop that fucking imperialist embargo against Cuba. |
- | Hasta Siempre Comandante Che Guevara. |
- | |
- | |
- | Fabio "Maverick" Bizzetti - bizzetti@mbox.vol.it - Maverick* at IRC |
- | The maker of "CyberMan" and "Virtual Karting" |
- | working on "Virtual Rally" and "StarFighter", the 3D game that will |
- | bring the Amiga to the top |
- | |
- ------------------------------------------------------------------------------
-
-
-