Kickflash experiences

[platon42]

@Tjaoz:
Ah, the little Elbox-Rat is back.

> It makes that the access to the Kickflash Flash memory is over 10 times slower than to eFlash 4000 Flash memory.

Oh, right! Except for that factor of ten must be from your wildest dreams, let's calculate: Reading 1MB over Zorro II with 2.8MB/sec is about 0.35 seconds. Now, reading 1MB over Zorro III with presumably 7MB/sec is 0.14 secs. Now this has saved you really some of your lifetime! My gosh!

> Paying a little bit more only you can have eFlash 4000 with 2MB(!) of Flash memory.

Yeah, and what would you want to fill in there? The eFlash software has no support for Kickstart flashing and it allows only a maximum of 15 modules. So first find something that makes sense to place into the flashrom before advertising with even more space. Unfortunately, the eFlash software does not support compression, unlike the Algor (Pro) software, hence, you could easily fit nearly same amount in an Algor Pro FlashRom-

In a german magazine, Amiga Plus (recent issue), Algor/Romulus, eFlash and KickFlash have been tested intensively. And guess what, the eFlash has the worst software (a "Installer script" to select the flash rom items with absolutely no user-friendliness), the worst documentation (less than 2 KB?), the least features (even normal romtags hardly work!) and least additional modules and goodies (there are none). But hey, at least the CD contains 41MB of Elbox website, with the software for the eFlash being less than 30KB.

I don't need to tell you what was the winning product in this review. Elbox' software is lightyears away from being something usable. Who cares about that Zorro III interface? The flashrom is only read ONCE after turning the machine one, who cares about a few milliseconds more or less.

[platon42]

orange wrote:
Maybe I didn't understand all this, but ROM is certainly read more than once unless it has been copied to RAM. All the programs use ROM routines, and most of Amiga OS is held in ROM  (thank God)
Can you explain me why are you all mentioning 1Gb, do you mean 1Mb? Kickflash is ROM image in flash memory, right?

There are only two "real" ROM spaces: 0xf80000-0xffffff where the normal kickstart ROM resides and 0xf00000-0xf7ffff for the "debugging" ROM (AFAIK used by CD32 and Cyberstorm/Blizzard early init ROM). The code in this location is already relocated to these fixed addresses -- they cannot be moved except by using virtual mapping by the MMU (or special hardware hacks like the MapRom feature).

Replacing Residents (aka RomTags aka Modules) from the ROM with new ones (e.g. from AmigaOS Update in 3.5/3.9/BB2) is done by relocating these modules somewhere in fast memory and letting a mechanism in the Kickstart ROM initialize the newer ones (KickTagPtr/KickMemPtr or the Residents list) instead the ones from the ROM.

Hence, as FlashROM actually behaves like a ROM in the sense that it cannot be written to in normal ways (it is not RAM), but you cannot guarantee that RomTags do not have DATA/BSS segments and need to write into associated memory or code, these modules have to be relocated to "real" RAM.

Under normal circumstances, this relocation of RomTags from the FlashRom only happens once at the very first boot, thus, the FlashRom is only read once and it is not time critical, how fast this is.

The Kickflash and Algor (PRO) cards are autobooting Zorro II cards, but they only use a 64KB segment of the Zorro II address space for accessing the 512KB/1MB of FlashRom.

Both Kickflash and Algor have to copy the contents to fast ram (including the kickstart rom image, that might be stored in there). (Again, this happens only once). The eFlash has to do this aswell, because of the reasons mentioned above (relocation) -- but unlike the other solutions, the eFlash cannot remap the Kickstart ROM.

The Kickflash has an additional clockport on its board, and it has been said that it should be able to hold a flash memory module with 1GB capacity (again, this will not show up as real memory in the computer, but rather some kind of ram-drive).

Hope this helps a bit.

[platon42]

@Tjaoz:

As you don't seem to get it into your head: Speed for reading the flashrom does not matter. It is not time critical. This is no video memory or data memory that has to be accessed all of the time. It is read once. Is a  Rat's brain too small to get this?

> The whole eFlash memory is directly available in the Amiga Zorro III area.

So what? It doesn't make it RAM. You cannot write it directly, so your point is moot.

> It opens the possibility of using the eFlash 4000 card not only during the computer start like other Flashes but also as the non-volatile Flash disk with fast random access.

So? The flashrom of the Algor is accessable at any time after the machine has booted -- how would you be able to reflash it otherwise? (AFAIK the flashrom of the kickflash can also be accessed after booting, so what are you talking about?) For your information -- there is a flashdisk.device for the Algor for more than a year now, which could be used for a non-volatile Flash disk.

> You should check your sources before posting your imagination numbers about Elbox products. Eflash 4000 software allows storing the AmigaOS ROM Update and 24 other modules. Not 15, as you wrote.

Oh, they must have updated their software then -- unfortunately, even 24 modules is not enough e.g. for all parts of Poseidon and the ROM Updates. And even specifying 24 modules on a command line in a shell is ridiculous! Is this what Elbox calls userfriendliness?

> So far software for eFlash 4000 does not need to use compression because this card includes from 2 to 4 times more Flash memory than other Amiga cards with Flash.

The Algor PRO has 1MB of flash memory. With compression, you get close too 2MB capacity. What would you like to store in there?

> AFAIK Elbox plans to include compression for OS4 modules.
Just like Elbox planned the Shark PPC, the eFlash1200 or the MPEG card drivers?

There is a difference between providing something and merely "planning".

> Adding the compression in the firmware is a matter of few hours for the software developer, though.

Sure.

> I read the comment from an OS4 betatester that there is no chance to insert the OS4 Kickstart files into 0.5MB of the Flash memory (Algor/Romulus) even if you use strong compression. So, no OS4 Kickstart files in Agor/Romulus Flash memory?

Unlike Elbox, E3B has an agreement about the Algor/Romulus flashrom being supported by OS4. I've been in contact with the Frieden brothers several times and they suppose it is possible.

> I read in someone's comment that in this Amiga Plus review eFlash 4000/1MB WON with Kickflash. But it looks that this is something, what you would rather prefer to hide here.

Your wildest dreams, huh?

The Algor was rated with 6/6,
the Romulus got 5,5/6,
the eFlash 5/6,
the KickFlash 4/6

I will never ever understand, why the eFlash still got 5/6 (must have something to do with Elbox advertising in the mag on a full cover page?) after reading the review, especially concerning the serious flaws of NO documentation, NO GUI except for an "Installer script" dialogue and BROKEN functionality (adding rom modules by hand didn't work). But I guess, this is because the card was ZorroIII instead of ZorroII, which makes absolutely no noticeable difference.

Ah yes, the Algor has won in this test BY FAR. This is nothing out of imagination. I really hope Total Amiga will do their own reviews soon enough. Advertisment and PR gags are one thing, real product quality another.

[platon42]

> Things like a way to add a security block to the system that's not easily overriddne by the early startup menu to mention one... another "lamer" thing would be to add sound on startup like the A1000 and other systems have...

Algor/Romulus software has this. The Sentinel software is a password protection system to lock your computer from unauthorized users. It can be either used synchroneously (the computer will stop booting until the correct password has been entered) or asynchroneously (it will continue to boot, though the screen is locked until you pass Sentinel.

Edit: Ah yes, the Algor also comes with modules to *disable* the bootmenu completely, if you like. This speeds up booting and protects you from the user selecting other booting devices.

Also, BlizKick modules like the A1000 startup sound are included. The Algor comes with an bootintro for the card, nothing keeps you from doing your own.

[platon42]

> Off topic (but then again, what was the topic here?).

It's not off-topic, I think.

>Does the Algor/Romulous really boot before the drives are spinned/scanned?

Yes, the Zorro cards are initialized straight after expansion and exec.library, hence this is one of the very first actions the Kickstart ROM does.

Hence, it could *also* add virtual drives from the FlashROM to the list of devices. In fact, the installation disk (880KB) is held in FlashROM as a module, which will pop up automatically on booting with the Algor inserted the first time (no other FlashROM solution has this).

[platon42]

> But what is the point in using Algor or Kickflash as ***a Flash disk***, if their read speed is lower than that of a hard disk?

The point is: There are several people that are using CF-IDE-Adapters that are not faster than aprox. 2MB/sec on the internal controller. Do they care about the speed? No. Do they care about the power and the noise? Yes.

The point is: Even the 1MB or 2MB of the eFlash is ridiculously small to hold a complete booting partition, so this point is moot. And hence, this 1 or 2MB of data will be read in at no time (it might even go into caches soon enough, just to compensate for your silly speed argument). So actually, having a filesystem on FlashROM of that size does not make any sense on *any* flashrom solution, regardless of the access speed. This might look different with the 1GB module for the KickFlash. There, it obviously makes sense to have a filesystem.

The point is: Elbox does not include any software for that theoretical flashdisk. Even their booting off their PCI USB card does not work, even though the card is advertised with that feature.

[platon42]

@tjaoz:
[off-topic BS skipped]

BTW: Whenever it comes to facts, you gladly seem to ignore them. Although this is no new observance, I find it nice of you to admit your defeat this way, over and over again :-D

>Here 16MB/s of the eFlash 4000 and full random access to any address of its one or two megabyte of the Flash memory go far beyond other Flash solutions.

So? Full random *read only* access is worth nothing. But of course you know this already.

> Isn't the Algor/Romulus memory access sequential only? Does it have to the 64kB of Flash memory be read before getting to the last word of each Algor/Romulus memory page?

You're talking BS again, little Rat. Why don't you simply admit that you don't have a single clue about how Flash memory works, especially on the Algor? The Algor has full random access (bytes, words, longwords) within a window of 4 KB. The Algor flashrom can be read at full Zorro II speed (2.8MB/sec).

> If yes, this would mean that average random access to Algor/Romulus Flash memory is at the level of ca. 88 kilobytes per second.

If you had a brain, this would mean that something went utterly wrong at some point of your life...

> As for now Algor Pro is a vapourware. Isn't it?

No, it isn't. It is in production (delayed by some health problems). Users already have recieved Luciferin V1.6, which fully supports the Algor PRO. Hence, the software is ready for weeks now, only the cards have to be shipped.

> If now OS4 Kickstart files prove to be too large for 0.5MB in Algor/Romulus Flash, will E3B make Frieden brothers guilty of that?

Another lame attempt to twist my words. I said that the Friedens and E3B have an agreement and they will keep it. It is none of your business, how it will be implemented.

> Looks like Elbox and Individual had long known that 0.5 MB is not enough.

For the eFlash board providing *only* a flashrom feature wasting a complete Zorro III slot, this is a very poor result (the KickFlash at least has got an additional clockport which the eFlash lacks). The Algor was developed AND released more than a year ago (!), and its flashrom capabilities were only a nice bonus (the main purpose was to provide a fast USB controller) -- and STILL its features and product quality are not beaten by other products, especially not by the eFlash, which comes with three command line tools only for basic reading and writing the flashrom and NO documentation.

As for the rest of your (off-topic) message: cut the crap. This is getting pathological. Every reader of the review will make up their minds themselves and will wonder about how the card still could get 5 out of 6 points.

Anyway, you never will acknowledge that compared to the other solutions, the eFlash clearly loses the competition, so any further arguments are in vain (as usually). And if you weren't stating your BS so obviously, people might be believing you more easily. But by now, everybody knows who pays you for this. :-P

[platon42]

Actually, I didn't want to feed the troll, but Rats seem to find their "food" anyway.

tjaoz wrote:
It is you [...]

Welcome to the kindergarden. "No!!! It's you!" as usually.

So? Full random *read only* access is worth nothing.

This is what you want others to believe. You know why?  Because random read access in E3B cards is so messed up that it is unusable.

Liar.

It is not true.

It is. Liar.

To make this speed possible, even in the quoted by you 4KB window range, at least 12 Amiga address lines should be connected to the Flash. This is not the fact!

It is. Liar.

Just watch the Algor board and see connection of address lines. Well, it is enough to read the description of the connector, to which Romulus is connected. This connector has only 5 Amiga address lines connected!

You have no idea. You're trying to tell Michael, who built the hardware and me, who wrote the software how the thing works? And you are talking about mental disorders? :laughing:

[Rest of stupidity skipped]

Algor pro IS vapourware, according to your own vapourware definition. Until today Algor pro is not available.

For the first eFlash 4000, Elbox took multiple months to actually deliver it -- even after they claimed, it would be available. Is there any customer of the eFlash 4000 2MB yet? I wonder.

The shipment Algor PRO boards has been delayed due to Michael going into hospital, which was officially announced right after this was clear. Elbox never ever posts any delays, it just silently doesn't deliver.

The quality of hardware of the Flash part of the Algor/Romulus cards is beaten by the Eflash 4000 hardware MANY times.

Argument and proof?

1. eFlash 4000 is a fast and clean design, while Algor implementation of the Flash memory is a hard hardware hack.

Why? Proof? Zero points. Anyone looking at the eFlash with some knowledge of hardware design will immediately notice that it is not a clean design at all. No need to repeat the facts Michael has given.

2. eFlash is equipped with hardware protection against unwanted reprogramming, which is very important because Flash memories have limited reprogramming cycles amount.

FlashRoms cannot be written directly, but need a special programming cycle, which cannot be triggered by accident. Hence, user interaction is necessary. Zero points.

When you have no hardware write protection (Algor and Romulus case) of the Flash memory you are always exposed to destroying your Flash memory by a malicious virus before you find out that some software is rewriting it again and again.

Your point is void. Each flashrom has to be written using a special vendor specific way. A virus would have to know exactly how to write itself into the flashrom for each different card. Also, just overwriting a block of the flashrom would not trigger it automatically. In case of the Algor/Highway, a checksum also has to be calculated (I'm not aware the eFlash has this). This makes it very hard to execute alien code, written into the flashrom accidentially or by malicious modification. And who would write such a special virus? (From Elbox' track record, I'd say, that with their reverse-engineering and trojan coding experience, this should be no problem for them). Zero points.

I have read in the review in the mag, that the eFlash can only be flashed with the hardware protection jumper disabled, but it will not execute the contents of the flashrom then -- now this pretty user unfriendly, if you only want to make quick tests.

3. eFlash 4000 uses hi-quality Flash memories, which can be programmed 100,000 times, not only 10,000 times like memories in Algor/Romulus.

The eFlash uses recycled and obsolete MACH chips that would start failing far before the flashroms reach their theoretical guaranteed flash times. These are minimum guaranteed times. Just calculate how many years you have to do constant flashing, until the *guaranteed* amount of cycles is reached? Zero points again.

I do not have to comment that the Algor's edge connector is not gold-plated, which in itself eliminates that product in the very beginning.

Why? Because before the connectors start wearing off (in 10 years?), the rest of the machine has fallen apart?
 

As regards your "fast" USB controller", it's a bit exaggeration. Algor is an old-fashioned obsolete USB 1.1 standard controller.

Getting off-topic again, huh? It reaches the same and faster transfer rates than the Spider with USB1.1 devices. 'Nuff said.

As a summary: All of your so-called "arguments" do not hold. Do cannot tell anything against my arguments that the software provided with the eFlash is a piece of sh*t. You cannot give any good reason, why one should favour a Zorro III board to a Zorro II board. Your hypothetical fast flashdisk is, well, hypothetical (and would be of no practical use on 1MB/2MB boards).

Better luck next time :laughing: