Amiga.org
Amiga News and Community Announcements => Amiga News and Community Announcements => Amiga Hardware News => Topic started by: DJS on July 30, 2003, 03:53:03 PM
-
Elbox Computer, the leading manufacturer of hardware for Amiga computers, announces promotion for all Mediator PCI busboards purchased in the Elbox Online Store from 30 July to 18 August 2003. With every Mediator busboard purchased in the promotion period, the end-user will be eligible to receive a free Terratec 512i Digital sound card enclosed in the Mediator shipment. For details, see the Special Offer page (http://www.buy.elbox.com/cgibin/shop?info=910M12S).
About Mediator:
The Mediator PCI busboard line is designed to enable expansion of Amiga computers with a wide range of PCI cards. The following PCI cards are already supported in Mediator busboards: Voodoo5, Voodoo4, Voodoo3, Voodoo Banshee, S3 VirgeDX and S3 Virge PCI graphic cards; Spider USB 2.0 host controller cards; Fast Ethernet 100Mbps and Ethernet 10Mbps PCI network cards; Sound Blaster 4.1 digital, Sound Blaster 128 and ForteMedia FM801-based sound cards; TV tuner cards. The Mediator busboard is a necessary expansion for Amiga computers to employ the power of the latest PowerPC G3/G4 processors, around which SharkPPC and SharkPPC+ cards are based. Mediator busboards are available for the following Amiga models:
- Amiga 4000T (Mediator PCI 3/4000T (http://www.buy.elbox.com/cgibin/shop?info=207M34T)),
- Amiga 4000D (Mediator PCI 4000Di (http://www.buy.elbox.com/cgibin/shop?info=203M4DI), Mediator PCI 4000D (http://www.buy.elbox.com/cgibin/shop?info=204M4D), Mediator PCI 3/4000T (http://www.buy.elbox.com/cgibin/shop?info=207M34T)),
- Amiga 3000T (Mediator PCI 3/4000T (http://www.buy.elbox.com/cgibin/shop?info=207M34T)),
- Amiga 3000D (Mediator PCI 3000D (http://www.buy.elbox.com/cgibin/shop?info=209M3D), Mediator PCI 3/4000T (http://www.buy.elbox.com/cgibin/shop?info=207M34T)),
- Amiga 1200 (Mediator PCI 1200 SX (http://www.buy.elbox.com/cgibin/shop?info=210M12S), Mediator PCI ZIV (http://www.buy.elbox.com/cgibin/shop?info=215M12Z4), Mediator PCI 1200 LT (http://www.buy.elbox.com/cgibin/shop?info=213M12LT)).
About Terratec 512i Digital:
The Terratec 512i Digital (http://www.buy.elbox.com/cgibin/shop?info=740TT512) offers everything you need to achieve perfect sound. With up to 4 channels for connecting 2 or 4 loudspeakers and top-quality recording and playback at up to 48 kHz you can enjoy crystal clear sound all around. TOS-Link digital output guarantees lossless recordings via direct optical connection for DAT or MiniDisc recorders. With 8- and 16-bit mono and stereo playback and 16-bit stereo recording plus user-selectable sampling rates from 5kHz to 48kHz, the Terratec 512i Digital produces precise audio fidelity.
Order your Mediator Now and get the best sound card for your Amiga for FREE!
-
eligible to receive?
-
I'd realy like to buy one of those Mediator 4000Di, but it would have been an better offer if they threw an VGA board into the offer instead of the soundcard. At least for me it would :)
-
Their website does not say "eligible to receive".
It says:
"A Terratec 512i Digital sound card for each end-user who purchases a Mediator PCI busboard from 30 July to 18 August 2003 in the Elbox Online Store. "
Loki :-)
-
Too bad they won't make an internal A3000 desktop like the 4000Di model. I may end up getting the "tower" A3000 one and have an ugly pile of hack on my desk, as I'm not going to towerize the thing at this point, but would still like a Mediator for it. Guess I just won't enjoy the nice 3000 desktop case anymore...
-
I see. Well poor choice of words in the news then.
-
And me'd like a version with the video slot. But guess that's not gonna happen either.
I've also seen some comments that the A3/4000 version doesn't have true DMA, like the A200 one. Can anyone confirm this?
-
"I've also seen some comments that the A3/4000 version doesn't have true DMA, like the A200 one. Can anyone confirm this?"
Why would you need a graphic card to make DMA transfers in a real DMA environment?
Moving the data to the gfx ram and then taking it with the CPU is a waste of time (and slower)... why do that if they had real DMA?
They didn't implemented DMA with their powerflyers Z3, as it's much more difficult to make it work without problems...
Afaik Mediators 3k work with old busters without DMA...
It's pretty clear for me that they did some changes to the 1200 version to speed up a little the bus and that's all...
On the other hand I have to say that prometheus uses the same method, so it doesn't have DMA, but the difference is that you have to send it to Poland and a Mediator can be upgraded by yourself.
The only card for A4000 with true DMA is the G-Rex (It's a pity that you must have a ppc to use it... what happens if your expensive ppc dies? ). The 4000T version of the G-Rex only has 3 slots... to few imho...
-
@Crumb
Mediator for A4000 has true DMA. From the Elbox Mediator PCI 4000 website:
---
• Busmastering/DMA in all PCI slots
Mediator PCI supports the busmastering mechanism in all its PCI slots. The PCI cards in the busmaster mode can offer extremely fast DMA access (up to 132MB/s)
without any CPU load. Up to five PCI busmaster cards may be used simultaneously in Mediator.
• SDRAM/SGRAM of PCI gfx card as a DMA buffer
Mediator PCI 4000 can use part of the graphic card memory for Amiga system needs. Thus, the entire system gains in effectiveness as graphic card's SDRAM/SGRAM
memory modules are about 10x faster than FastRAM memory in Amiga turbo cards. With very short access time in SDRAM/SGRAM, several PCI busmaster cards may use this memory simultaneously. The PCI cards may then rush at the data transfer rate of up
to 132MB/s without any CPU load whatsoever.
• DMA to A3/4000 motherboard space
Mediator PCI 4000 includes a busmastering controller supporting transfers between PCI cards and the A3/4000 motherboard space.
---
Due to the A4000 design limitation DMA to A3/4000 motherboard cannot reach 132MB/s which is required by PCI specification for such transfers. It is why Elbox has not used this transfer method for the busmaster PCI cards. Elbox has implemented DMA to the A4000 motherboard in Mediator 4000 for the other purposes. This mechanism is prepared to allow SkarkPPC cards to access the A4000 motherboard registers. Mediator PCI 4000 has the jumper named MASTER for this mode activation.
AFAIK G-Rex forces the PCI busmaster cards to do DMA to the turbo card memory but this mechanism in Grex is much too slow for problem-free work. It is why Grex boards have so many problems with the simultaneous operation of more than one PCI busmaster card in the system. This is probably one of the reasons why there is only so few drivers for Grex. There is not sense to develop drivers for new PCI cards if they cannot co-operate with others.
-
@tjaoz
-• Busmastering/DMA in all PCI slots
sorry, but that DMA ultrafast access is only between PCI cards, not between your mediator and the motherboard, and that's the problem.ç
-• SDRAM/SGRAM of PCI gfx card as a DMA buffer
that's a kludge, no one would need a DMA buffer if it had real DMA access to the motherboard. What they don't tell you is that the cpu moves slooowly that data from the gfx ram to the fast ram.
• DMA to A3/4000 motherboard space
muaahahah :-D
good try... old busters like 6-7 don't have Zorro3 DMA implemented and mediators 3k work as they do with Buster11
"Due to the A4000 design limitation DMA to A3/4000 motherboard cannot reach 132MB/s which is required by PCI specification for such transfers. It is why Elbox has not used this transfer method for the busmaster PCI cards."
That sounds better, but Busters6-7 work without problems with Mediators and DMA support wasn't added to Zorro3 until Buster9 was released.
And I don't believe that PCI specification forces you to write to ram at 132MB... on a 486 with PCIs if other DMA channels are accessing Ram you probably saturate the PCI bus and the data will be written in ram slowly...
I'm not much interested in vapourware cards... doing something like making the design worse than it could be to recycle the 1200 design is simply crap. Zorro3 should give you a bandwitch of 16MB/s with a gfx card (like a CV64 not 3D), not around 10MB (on the 4k version) like now with a Mediator4k or a Picasso4...
"AFAIK G-Rex forces the PCI busmaster cards to do DMA to the turbo card memory but this mechanism in Grex is much too slow for problem-free work. It is why Grex boards have so many problems with the simultaneous operation of more than one PCI busmaster card in the system. This is probably one of the reasons why there is only so few drivers for Grex. There is not sense to develop drivers for new PCI cards if they cannot co-operate with others."
first of all, I don't like the idea of needing a ppc to have proper pcis, but sorry, G-Rex HAS lots of drivers: TV, Terratec sound, Voodoo, ethernet 100, ethernet 10,even USB (unreleased)...
And ok, the 1200 version is crap, but the 4000 version seems to work really well.
BTW it also has OpenPCI support and it seems that when a Mediator version of this library was going to be released the DMA stuff worked much better with G-Rex...
-
G-Rex HAS lots of drivers: TV, Terratec sound, Voodoo, ethernet 100, ethernet 10,even USB (unreleased)...
Are you drunk? Mediator has a lot of drivers for Tv cards (over 80-90), for voodoo blaster 3, 4, 5, ethernet 10/100, Sbblaster and Terratec. Plus USB fast 2.0 !!!
Mediator driver is always thefirs to be released, and the G-Rex perfomance in the 1200 aren't very well.
Also there are a lot of drivers like mpeg decodes for Mediator that are going to released.
Sorry But G-Rex is a "flop", a very bad products and very silly. It didn't sell well, and it required a stupid olt PowerUp. Phase Phase was good, DC really sucks. Mediator is a good piece of hardae, with fast drivers. Prometheus are a good board but not a real solution, because it is only Zorro based.
-
@ikir:
"Are you drunk? Mediator has a lot of drivers for Tv cards "
of course I haven't said that mediator has not drivers what I have said is that G-Rex has lots of drivers too... if the USB drivers are released they will have almost the same number...
"Mediator driver is always thefirs to be released, "
BTW G-Rex users had Terratecs playing music long time before Mediator users ;-)
And If I remember correctly they had Voodoo5 CGX drivers before Mediator.
Their TV program is truly skinable, have you seen it? I'd like to know how it is programmed...
"Also there are a lot of drivers like mpeg decodes for Mediator that are going to released."
Vapourware at the moment.
"Prometheus are a good board but not a real solution, because it is only Zorro based."
Excuse me? The 3000/4000 versions of the mediator connect are connected to the Zorro3 bus... The only thing against it that I have is that it's difficult to update and that it fits bad in my A4000T...
"Mediator is a good piece of hardae, with fast drivers. "
I'd like to see some benchmarks ;-) Their Ethernet drivers seems to be quite slow...
"Sorry But G-Rex is a "flop", a very bad products and very silly."
If we talk about the 4000(/T) version, I don't agree. It seems to be faster than Mediator (at least the gfx card...)
I guess that when MorphOS is released for Cyberstorms we will be able to compare it to OS4+Cybestorm and see what of the two is faster/has more features... for example I think that the performance of their crappy DMA hack is worse than Elbox thought initially and that's the reason they don't release SCSI/IDE drivers... A Vapourware-SharkPPC probably would make all these Mediator problems dissapear... but with current hardware their implementation seems a little crap.
BTW: When will be released the linux drivers? I guess we won't see MorphOS drivers, it's a pity, I'd like to buy a pci bus that fits well in my A4000T and that allows me to run OS4 and MorphOS... Prometheus fits quite bad, so it's not an option...
OS4 will not have G-Rex drivers, so I guess my only option would be the Mediator4kT :-/
-
@Crumb
that's a kludge, no one would need a DMA buffer if it had real DMA access to the motherboard.
No. DMA access to the A4000 motherboard is too slow for PCI cards.
BTW DMA is not needed for operation of some cards at all: Voodoo and Ethernet 10Mbps never use DMA.
What they don't tell you is that the cpu moves slooowly that data from the gfx ram to the fast ram.
CPU sends data from the graphic card memory to fast ram as fast as the turbo/processor card allows. Some turbo cards do it faster (e.g. Apollo cards), other perform slower (e.g. Blizzard cards). This speed has nothing to do with Mediator. Mediator is prepared for much higher
speeds, up to 132MB/s.
good try... old busters like 6-7 don't have Zorro3 DMA implemented and mediators 3k work as they do with Buster11
What is a problem? It is obvious that if you would like to use DMA to the motherboard you need to update the buster to rev.9 as a minimum.
Zorro3 should give you a bandwitch of 16MB/s with a gfx card (like a CV64 not 3D), not around 10MB (on the 4k version) like now with a Mediator4k or a Picasso4...
This bandwidth depends on the processor card, not on Mediator.
first of all, I don't like the idea of needing a ppc to have proper pcis, but sorry, G-Rex HAS lots of drivers: TV, Terratec sound, Voodoo, ethernet 100, ethernet 10,
I have never heard that anybody has more than two stable operating DMA cards with a Grex.
So, drivers for them are useless ;-)
even USB (unreleased)...
USB for Grex is vapourware.
BTW it also has OpenPCI support and it seems that when a Mediator version of this library was going to be released the DMA stuff worked much better with G-Rex...
OpenPCI is nothing interesting for Mediator users. Mediator has MANY more drivers than other PCI solutions. See here (http://www.elbox.com/mdg.html) :-D
-
@Crumb
BTW G-Rex users had Terratecs playing music long time before Mediator users
BTW. Mediator users had SB128 cards playing music long time before Grex users had sound drivers at all. :-)
And If I remember correctly they had Voodoo5 CGX drivers before Mediator.
Mediator users used Voodoo5 when Grex was not there at all :-D
Their TV program is truly skinable, have you seen it? I'd like to know how it is programmed...
Mediator has many excellent TV programs. The best are SuperTV and TvR.
"Prometheus are a good board but not a real solution, because it is only Zorro based."
Excuse me? The 3000/4000 versions of the mediator connect are connected to the Zorro3 bus... The only thing against it that I have is that it's difficult to update and that it fits bad in my A4000T...
Prometheus is not supported any more. The guy who designed it withdrew from this project altogether. He even sold his own Prometheus recently.
"Mediator is a good piece of hardae, with fast drivers. " I'd like to see some benchmarks Their Ethernet drivers seems to be quite slow...
Slow? Their drivers are excellent and very fast. I had about 950kB/s with an Ethernet card.
for example I think that the performance of their crappy DMA hack is worse than Elbox thought initially and that's the reason they don't release SCSI/IDE drivers... A Vapourware-SharkPPC probably would make all these Mediator problems dissapear... but with current hardware their implementation seems a little crap.
In my opinion the DMA gfx buffer idea developed by them is an excellent solution to bypass limitations of present-day turbo cards.
If you really need to call something a "crappy DMA hack," please reserve this for something like April chips for the ArticiaS. Here it is justified if using April chips does not solve all DMA problems with the ArticiaS.
OS4 will not have G-Rex drivers, so I guess my only option would be the Mediator4kT :-/
A good choice!
-
"No. DMA access to the A4000 motherboard is too slow for PCI cards."
Excuse me but Zorro3 scsi cards use DMA without problems the bandwitch is more than enough to develope scsi drivers or ethernet100 drivers that write directly to the fast ram.
Every data you want to take from the mediator and put in the Fast Ram as to pass trough the Zorro3 bus. With a cpu you have to move everything by yourself and depending on the load of the cpu you may not get 100% of the bandwitch, but with DMA you will get more speed than using the CPU because you can write in fast ram/Zorro3 memory area at the maximum speed your memory allows.
BTW can you prove that using the cpu to move data from a ZorroIII memory area to fastram is faster using the cpu than using DMA? If you plug some ram in a Zorro3 ram card you should get nearly the maximum speed of Zorro3, and with a buffers from a harddisk for example created in that memory you wouldn't be able to get more speed writting and reading using the cpu... the maximum speed you can read/write to the PCI gfx ram depends a lot on the mediator. If it had a better design that allowed you to read/write at 16MB/s, then the current design would be somewhat strange but it would be OK. The problem is that the speed is limited by the mediator to ~9MB/s (remember that I'm talking about A3000/4000 version, and they work in a slightly different way than A1200 versions...)
In conclusion:
how can you say that the Mediator4k is well implemented when you can write to a CV64 (not 3D) at 16MB/s and a Mediator4k in the same machine with a Voodoo5 only gives you ~9MB/s?
"CPU sends data from the graphic card memory to fast ram as fast as the turbo/processor card allows. [...] This speed has nothing to do with Mediator. Mediator is prepared for much higher
speeds, up to 132MB/s."
Sorry, that speed is only achieved between PCI slots, and we are talking about the A4000 version, so the design of the mediator as a lot to do with speed... It works as a ZorroIII card and for your information not all ZorroIII cards work as well, you have excellent implementations like CV64 (not 3D) and crappy ones like CV3D. You can use DMA like in a FastLane Z3 or you can use the CPU like in a PoweflyerZ4. There are different ways of making a ZorroIII card and Mediator3000/4000 is far from being optimal...
"What is a problem? It is obvious that if you would like to use DMA to the motherboard you need to update the buster to rev.9 as a minimum."
The problem is that it gets exactly the same speed because every data movement from the mediator to the fastram is done by the cpu, just because it doesn't do DMA between the Mediator and the FastRam... That's the reason the Buster doesn't matter
"Quote:
--------------------------------------------------------------------------------
Zorro3 should give you a bandwitch of 16MB/s with a gfx card (like a CV64 not 3D), not around 10MB (on the 4k version) like now with a Mediator4k or a Picasso4...
--------------------------------------------------------------------------------
This bandwidth depends on the processor card, not on Mediator."
In the 1200 version yes, in the 4000 version NO. In the 4000 version Mediator is no more than a Z3 card and mediator's design affects a lot to the performance of PCI cards.
"I have never heard that anybody has more than two stable operating DMA cards with a Grex.
So, drivers for them are useless "
Maybe you are talking about the 1200 version, which only has 2 DMA slots. In the G-Rex 4000 all slots have DMA.
"USB for Grex is vapourware."
at the moment yes.
"OpenPCI is nothing interesting for Mediator users. Mediator has MANY more drivers than other PCI solutions. See here "
Voodoo3/4/5, Permedia2, SiS, BT878/848, SB128, Terratec,Ethernet10/Ethernet100,etc... The only driver Mediator has and G-Rex has not is USB.
Anyway you can't register Poseidon if you buy a Spider now, so it's a card without drivers unless you are a pirate.
So I guess I can say that the Spider is "vapourware" because you have not (legal) way to make it work (or you won't have in 10 days if Elbox doesn't come to an agreement with Chris Hodges).
LOL :-D
-
"Mediator users used Voodoo5 when Grex was not there at all "
I doubt that because the first Voodoo5 drivers were CGX, see old vgr news archive...
"Prometheus is not supported any more. The guy who designed it withdrew from this project altogether. He even sold his own Prometheus recently."
It is supported by OpenPCI.library, as the Prometheus is well documented the drivers developed in other OpenPCI.library compliant systems work without modifications. And the best of all is that these drivers are free and work in almost every Amiga platform (Pegasos/Amithlon/Prometheus/G-Rex) It was stupid from Elbox to force Titan not to release it... they didn't want people saw that they can make work an Ethernet100 without buying their MMCD...
"Slow? Their drivers are excellent and very fast. I had about 950kB/s with an Ethernet card."
Pretty poor for an Ethernet100
"In my opinion the DMA gfx buffer idea developed by them is an excellent solution to bypass limitations of present-day turbo cards."
In mine opinion it is a hack, but it is crappy because you get less speed from a PCI gfx card than from a well made Zorro3 gfx card.
-
I totally agree with tjaoz, he explain the thing perfectly.
i like elbox very much and i support they... the only thing tha i don't like is the Poseidon/OS4/Picasso96 Elbox politics.
I'll buy a spider and a shark only it they support theri author, and these products will have OFFICIAL drivers.
-
"""Pretty poor for an Ethernet100""""
Slow? Are you drunk again? :-o Do you confuse Mbit and Mbyte?
-
tjaoz (Rat) and you aren't siamese twins by any chance? Because you always say the same.
-
For 100MBit Ethernet, that normally does about 7 MegaByte per second on any PC, yes, 900KB is really poor. My HydraNexus (10MBit) did about 700KB with AmiTCP.
-
About the Articia stuff... You don't really know what
you're talking about, do you? You know about one single
bug and think that every bug is incarnations of this...
-
My network card on my 1200 works great, considering that i have only a 060.
So i think you must support users Platoon not insult them.
-
@Crumb
Excuse me but Zorro3 scsi cards use DMA without problems the bandwitch is more than enough to develope scsi drivers or ethernet100 drivers that write directly to the fast ram.
The fact that DMA access speed to the A4000 motherboard is sufficient for Zorro cards working in the DMA mode does not mean it is sufficient for PCI cards in the DMA mode.
BTW can you prove that using the cpu to move data from a ZorroIII memory area to fastram is faster using the cpu than using DMA?
It depends on the situation.
When many small data packets are transmitted over USB or over a network the USB controller or the FastEthernet card collect these packets and store them in a large buffer in the graphic card memory without loading the Amiga system at all. When the buffer is full, the processor copies the entire buffer in one go. With this mechanism, effective transmission speed may be higher than in the situation of a PCI card initiating transmission of DMA to the computer’s memory for each data packet.
The problem is that the speed is limited by the mediator to ~9MB/s (remember that I'm talking about A3000/4000 version, and they work in a slightly different way than A1200 versions...)
AFAIK this speed is usually rather 10-12MB/s not 9MB/s and it depends on the A3/4000 processor’s card model. I saw a bustest of A3000 with Mediator and Voodoo3. The long words write speed was over 11MB/s.
Maybe you are talking about the 1200 version, which only has 2 DMA slots. In the G-Rex 4000 all slots have DMA.
No. I am speaking about all G-rex versions. No more than two DMA cards can operate with Grex 4000 in a stable way. In Grex 1200 there is only one or two (depending on version) slots in which DMA PCI cards can be fitted. More DMA PCI cards cannot be used at all.
The only driver Mediator has and G-Rex has not is USB.
AFAIK OpenPCI Fast Ethernet driver does not work with Grex (it does on Amithlon only) and Grex do not have SB 128 drivers. You can find a comparison between PCI cards supported in Amiga PCI busboards here (http://elbox.com/tests/comparison_en.html). It is from 1 October 2002.
Anyway you can't register Poseidon if you buy a Spider now,
I bought Spider last year :-)
So I guess I can say that the Spider is "vapourware" because you have not (legal) way to make it work (or you won't have in 10 days if Elbox doesn't come to an agreement with Chris Hodges).
Elbox has had an agreement with Hodges since September 2002. See here (http://www.amiga.org/forums/showthread.php?t=43332) and here (http://groups.yahoo.com/group/Amiga-Mediator/message/13629).
-
@Crumb
I doubt that because the first Voodoo5 drivers were CGX, see old vgr news archive...
In the vgr news site, many products and drivers appear, which are pure vapourware, like USB cards and drivers for Grex.
It is supported by OpenPCI.library, as the Prometheus is well documented the drivers developed in other OpenPCI.library compliant systems work without modifications.
I have not heard of anybody with these OpenPCI drivers working with the Prometheus.
Prometheus does not have DMA to A3/4000 motherboard and does not have working DMA between PCI slots. When the Matay company existed they informed about the firmware update but this update has never worked properly, AFAIK.
It was stupid from Elbox to force Titan not to release it...
Titan received Mediator SDK and Elbox pci.library documentation to write drivers for Mediator, and not to write his own pci library.
they didn't want people saw that they can make work an Ethernet100 without buying their MMCD...
What? So why Titan has NOT done these drivers for Mediator using Elbox pci.library? That was the purpose for which he received Mediator SDK from Elbox.
"Slow? Their drivers are excellent and very fast. I had about 950kB/s with an Ethernet card."
Pretty poor for an Ethernet100
I wrote about Ethernet (10Mbps) card, not FastEthernet (100Mbps). Can't you see the difference?
"In my opinion the DMA gfx buffer idea developed by them is an excellent solution to bypass limitations of present-day turbo cards."
In mine opinion it is a hack, but it is crappy because you get less speed from a PCI gfx card than from a well made Zorro3 gfx card.
Sorry? What does PCI gfx card speed have to do with the DMA gfx buffer solution?
Voodoo gfx card does not work in DMA.
-
@platon42
tjaoz (Rat) and you aren't siamese twins by any chance? Because you always say the same.
platon42 (Hodges) and you aren't siamese twins by any chance? Because you always say the same.
900KB is really poor. My HydraNexus (10MBit) did about 700KB with AmiTCP.
It is rather your HydraNexus, which is poor. As I wrote my Ethernet (10Mbps) NE2000 card in Mediator runs at 900KB.
-
@AmiGR
About the Articia stuff... You don't really know what
you're talking about, do you? You know about one single
bug and think that every bug is incarnations of this...
If you are so knowledgeable, why don't you tell us more about these bugs in Pegasos/April2 hardware?
-
Hi rat,
always a pleasure to read your comments.
It is rather your HydraNexus, which is poor. As I wrote my Ethernet (10Mbps) NE2000 card in Mediator runs at 900KB.
So, as I understand you, a rather old network card running on a slow Zorro II bus reaches 78% speed of your superior PCI solution with nowadays PCI network cards... and you call this "poor" ? Wuha.
Michael
-
"Slow? Are you drunk again? Do you confuse Mbit and Mbyte?"
I can get 5-6MEGABYTES/Second with a crappy 8139 in my peecee. With a crappy pentium100 and a crappy 8029 I got 1MEGABYTE/Second without problems
-
@tjaoz:
Do you work for Elbox? :-P
I can't see how can you defend them... OpenPCI only means more drivers in the future. In the software side Elbox sucks. They have only caused problems. Firstly problems with the CGX team, then problems with the P96 team, then problems with Chris Hodges, who's next? Let software developers do they work and concentrate in your hardware stuff.