»

[AmiBoy]

Hi Everyone,

I looking at replacing my bog standard buffered IDEFIx interface with a "much faster" IDefix express or Powerflyer one but before I buy I would like to know which one is actually better (i.e. faster speeds/lower CPU/memmory usage etc).

I would like to stay with using IDEFix (so the Express verison is s good option for me) as I have used it for many years but if the powerflyer is much better then I may go for that!

Thanks for any replies!

AmiBoy

[AmiBoy]

BUMP

[amigakit]

Powerflyer (?) dont you mean FastATA ;-)

Seriously though, when this was discussed recently here on Amiga.org, even though the FastATA is technically faster, in real world situations, speeds that were quoted from the IDE-FIX Express were suprisingly very good.

[AmigaMance]

Hello.
I read in the Amigakit website that "Transfer rates of more than 5.1 MBytes per second are reached with the express adapter".
How much faster is that from the standard IDE port of the Amiga 1200?

[keropi]

with the new mkIII fastatas, you can get more than 6mb/sec...
I have the MK1 32bit one (the older 1st 32bit version) and I get 6.1MB/sec , so the newers can outperfom that with less cpu usage...
also the FastATA driver is very good, no need to worry if u cannot use idefix97.
AFAIK the standar a1200 ide is 500-600kb/sec...

[darksun9210]

with good or new drives, the standard A1200 IDE port can push around 2.5Mb/s...
off the powerfyler / FastATA i've got over 5.6Mb a second... but i'm interested to see what the ver8.4 drivers can do as the ones i have are little "weird" to say the least.
thinking of ditching IDE and going scsi with a couple of scsi2ide drive adapters. 8-10Mb/s DMA with virtually no CPU load and no driver issues is quite tempting...

[humppa]

I read in the Amigakit website that "Transfer rates of more than 5.1 MBytes per second are reached with the express adapter". How much faster is that from the standard IDE port of the Amiga 1200?

Well, 5.1 MB/s is about 2x of what you get from a standard (non-Express) adapter. You will be a lucky guy if you manage to achieve these transfer rates with the Express-adapter.
The actual speed you get from it strongly depends on your accelerator. Using an Apollo 1240/40 I nearly achieved 5 MB/s. Then I switched to a BPPC 175/040/25 and the transfer rates dropped significantly.
Then I finally made the switch to SCSI: 7.8MB/s with very low CPU utilization, not a completely hogged Miggy (as with the Powerflyer).

[Mad-Matt]

Internal ide will even hit its 2.9-3.0mb/s with a fast 030 processor too(with a little help from idefix).

[tjaoz]

IDE-Fix Express is comparable in speed with the PowerFlyer Junior. Both were reviewed in Amiga Format. You can read this review here. PowerFlyer Junior is a 16-bit version of the A1200 PowerFlyer (aka FastATA 1200) controller. 32-bit version is about twice faster in the same config.

[tjaoz]

@humppa

Using an Apollo 1240/40 I nearly achieved 5 MB/s.

In the same config with FastATA instead of IDE-Fix Express you should have about 10MB/s. Here is a page with the FastATA 1200 test results.

[humppa]

@tjaoz

As can be seen from Sysinfo, they overclocked the 68040 to 44.4 Mhz in order to have transfers more synchronous which further speeds up transfers.
The Sysinfo drive benchmark is buggy anyway and you can't use it for comparison.

DiskSpeed 4.2 results (at the bottom) are more realistic. Anyway, I would always prefer SCSI. I don't care about fast transfers as long as MP3 playback gets choppy each time I copy a larger file. With SCSI you won't have any of these problems.

Moreover, it's not only about speed. The Idefix Express is less of a hassle to fit, especially when using a Mediator. The Powerflyer is much more unreliable.

[tjaoz]

@humppa

The Sysinfo drive benchmark is buggy anyway and you can't use it for comparison.
DiskSpeed 4.2 results (at the bottom) are more realistic.

No. The Sysinfo drive benchmarks are not buggy.

It looks that you do not understand what is tested by different benchmark programs:

1. Sysinfo and DriveSpeed measure disk raw speed. These programs are the best to test how fast the IDE controller transfers raw data from the hard drive.

2. DiskSpeed measures the speed of reading and writing data from/to the file. DiskSpeed results highly depend on the file system which is installed on your hard drive.

You cannot compare apples to oranges. I mean you cannot compare Sysinfo IDE-Fix Express results with DiskSpeed FastATA results, or to compare the results from systems equipped with the different turbo cards.

The Idefix Express is less of a hassle to fit, especially when using a Mediator.

I have FastATA under the Mediator board and I did not have any problem with installation.

[amigakit]

I am using a FastATA here and if it is firmly installed, it will give no problems.

[humppa]

No. The Sysinfo drive benchmarks are not buggy.

They are not reliable. When I had the Idefix Express connected, my Seagate drive always reported something around 5MB/s, the values kept on changing each time I measured, everything ok there. The other drive (Quantum) always reported exactly the same value of about 7.8MB/s (!): A value that is not realistic for the device. Further, it is very unlikely that the raw speed always exactly stays at the same byte without any fluctuation.
I have heard similar reports from other users. The speed-test in Sysinfo is a nice add-on, but I would always prefer dedicated programs which measure more than just raw speed.

In the end we are interested in real-life performance, not something like those buggy values I got from Sysinfo.

You cannot compare apples to oranges. I mean you cannot compare Sysinfo IDE-Fix Express results with DiskSpeed FastATA results, or to compare the results from systems equipped with the different turbo cards.

Congrats. But you started comparing an Apollo clocked at 44.4Mhz with SysInfo benchmark to a number given by me for an undefined program and a clock speed of 40Mhz. So who actually compared apples to oranges?

I have FastATA under the Mediator board and I did not have any problem with installation.

Fine. I did not say that it won't work. I just have heard many reports of people who had problems with the installation and who had problems after the installation because the whole construction was rather unstable.

[tjaoz]

@humppa

They are not reliable. When I had the Idefix Express connected, my Seagate drive always reported something around 5MB/s, the values kept on changing each time I measured, everything ok there. The other drive (Quantum) always reported exactly the same value of about 7.8MB/s (!): A value that is not realistic for the device. Further, it is very unlikely that the raw speed always exactly stays at the same byte without any fluctuation.

Even making simple tests requires some level of skills. :roll:

If you want to obtain reliable results from programs like SysInfo or DriveSpeed you should perform tests when your system works without any patches. The good habit is to boot a computer from a clean OS partition for any tests. Before starting a benchmark program, start Setpatch and the driver for the controller (ATA3.driver in case of FastATA and IdeFix in case of IDE-Fix Express) only.

The speed-test in Sysinfo is a nice add-on, but I would always prefer dedicated programs which measure more than just raw speed.

In the end we are interested in real-life performance, not something like those buggy values I got from Sysinfo.

We are talking here about performance offered by two pieces of hardware: FastATA 1200 and IDE-Fix Express, not about how slow your computer is "in real-life performance" due to improper or not optimum configuration of your software. To compare performance of the hard drive controllers you should use test programs which test raw speed, not programs (like DiskSpeed) whose results heavily depend on the filesystem (OFS, FFS, SFS, PFS3) installed on the hard drive.

Congrats. But you started comparing an Apollo clocked at 44.4Mhz with SysInfo benchmark to a number given by me for an undefined program and a clock speed of 40Mhz. So who actually compared apples to oranges?

This is not true. Please read once again what I wrote. You wrote that using an Apollo 1240/40 with IDE-Fix Express you achieved nearly 5 MB/s, and I answered that in the same config with FastATA instead of IDE-Fix Express you should have about 10 MB/s. I had Apollo 1240/40 (not overclocked) and my hard drives achieved about 10 MB/s.

It is obvious that performance of FastATA is twice better than the IDE-Fix Express performance. It results from differences in their design.

Speed of any fast IDE controller used in Amiga 1200 is limited by the design of the A1200 turbo card. The FastATA 1200 controller accesses a hard drive at the maximum speed with which the particular A1200 turbo card can read/write data to/from A1200 motherboard. IDE-Fix Express cannot operate faster than 1/2 of this speed. It is because the FastATA 1200 is connected via the 32-bit data bus, and IDE-Fix Express is connected via the 16-bit data bus only. So in each transfer from/to the turbo card the FastATA receives/sends twice more data than IDE-Fix Express. Period.