I have an Asantefast 10/100 Nubus card installed on a Quadra 700 and I can only get about 225 KB/s download speeds. The lights on the card are both green showing it is connected at 100Base-T speeds. The 225 KB/s is basically the same speed I get with the onboard 10Base-T ethernet connection. I would expect it to be significantly faster. Any ideas why this would be?
-
Hello MLAers! We've re-enabled auto-approval for accounts. If you are still waiting on account approval, please check this thread for more information.
Asantefast 10/100 Nubus Ethernet on Quadra 700 - Slow Speed
- Thread starter Macunix
- Start date
My test rig is NetGear Gigabit switch (GS105) connected to a Macbook Pro 2020 (1 Gig Ethernet adapter) running Apache. I plug the Quadra 700 directly into the same switch. I am using three foot long Cat6 ethernet cables for the connection and Netscape Communicator to download a test file. The switch shows 1000M connection for the Macbook Pro 2020 and 100M connection for the Quadra 700. The correct drivers are installed, although looking at the extensions folder, I don't see one labeled Asante in any way. For comparison, my Powerbook G4 running OS9 and Netscape Communicator downloads the same file at 5100 KB/s when connected to the same switch.
In addition to a PowerBook G4 being ~10-50x faster (minimum) than a Quadra 700, Classic Mac OS is just slow at networking, and most software can't saturate the existing connections on these machines. Your PowerBook G4 will probably move faster on the network if you run OS X on it, for example.
Here's some benches somebody got a couple years ago, where they achieved a "noticeable" speed boost from ~3 to ~4 megabits/second.
68kmla.org
The whole thread is interesting, but the TL;DR is that this card doesn't do much in most normal circumstances. If I had to guess, it's "for" high frequency AWGS8150 and 9150 machines using software that matches things up with client software (like installing the network version of Speed Doubler on both sides) or "for" high-speed 8100s doing specialty work with NuBus cards that weren't replicated in PCI in time to move the work to an 8500 or 9500.
Tough to say though because even with a G3/400 the tests in the other thread barely broke 10 megabits under like one or two trials.
That said, the 10/100 card did go faster and I don't really know why.
Here's some benches somebody got a couple years ago, where they achieved a "noticeable" speed boost from ~3 to ~4 megabits/second.
68K/early PPC with onboard AAUI 100mbps ethernet?
I learned a lot from this thread, but it looks like nobody ever posted test results... I have some here for those still curious: Test file: Mac OS 9.2.1 Update.smi (84.4MB / 86,018,641 bytes) Switch: TrendNET TEG-S24g Gigabit Server: Power Macintosh 9600/1000(G4 [2MB Cache on board]) Apple...
68kmla.org
The whole thread is interesting, but the TL;DR is that this card doesn't do much in most normal circumstances. If I had to guess, it's "for" high frequency AWGS8150 and 9150 machines using software that matches things up with client software (like installing the network version of Speed Doubler on both sides) or "for" high-speed 8100s doing specialty work with NuBus cards that weren't replicated in PCI in time to move the work to an 8500 or 9500.
Tough to say though because even with a G3/400 the tests in the other thread barely broke 10 megabits under like one or two trials.
That said, the 10/100 card did go faster and I don't really know why.
Wasn't there something about different versions of networking software (specifically MacTCP vs. OpenTransport) wildly affecting transfer speeds? There was no mention of what OS is in use here. Maybe try benchmarking under different OSes, specifically OS 8.1, to see if there are any differences.
I'd speculate that the bulk of the 100M speeds provided in these NuBus cards was more to prevent network speed mismatch hiccups on a hub-based network or something than it was to actually give blazing transfer speeds.
I'd speculate that the bulk of the 100M speeds provided in these NuBus cards was more to prevent network speed mismatch hiccups on a hub-based network or something than it was to actually give blazing transfer speeds.
Besides classic MacOS being slow at networking, it could just be a lousy application of the Ethernet chip used. The SMC FEAST (what a weird name) supports both 16-bit and 32-bit bus applications. The datasheet gives examples for ISA (16bit), EISA, and VLB (32bit) operation. Asante might have gone the 16-bit route with NuBus if they recycled an earlier 10Mbit Ethernet card design as a base. It's doubtful the card was designed for burst transfers or anything like DMA, likely just polled I/O.
That being said, I'd love to see these benchmarks with the 10/100 card re-run on a x100 PPC machine (which should have proper NuBus 90 support) with both MacOS and some sort of *NIX. Same for 68k given that somewhat updated *NIX systems still exist for them.
That being said, I'd love to see these benchmarks with the 10/100 card re-run on a x100 PPC machine (which should have proper NuBus 90 support) with both MacOS and some sort of *NIX. Same for 68k given that somewhat updated *NIX systems still exist for them.
Well I finally got around to testing the card under NetBSD 9.2 on my Quadra 700. The speed was about 340 KB/s, so its a little faster than under MacOS 8.1 with Open Transport (225 KB/s). I used the same test setup as discussed in my earlier post. I think the architecture is the bottleneck for this card. I was impressed with the fact that NetBSD had drivers for the card.
