SiliconExpress IV vs 840av internal SCSI? w/SCSI2SDv6

[jeremywork]

SCSI2SDv6 on internal 840av bus (bottom line)

vs

IBM 17gb 10k SCSI (Hitachi) on SEIV bus (top line)

FYI the SCSI2SDv6 is using a slower 6.2.5 firmware, as newer firmware cause corruption on write at the moment (is being fixed)

View attachment 30920

I took a longer look at this picture and have a couple more observations. My 840av is out for the count so I can't directly compare, but that internal bus reading looks too low- either the SCSI2SD will have no benefit in the SEIV, or there's some other bottleneck here. I get over 3MB/s on a period 230MB hard drive on a Quadra 700, and a bit over 4MB/s for a later 1.2GB drive on a 950. The 840av having a faster bus should put it even closer to the 5MB/s Synchronous limit of its SCSI controller.

On the other hand, your SEIV is much faster than I can achieve in my 950. I can only get around 8MB/s on a 15k SCA>68pin, but that's almost all due to lack of full Nubus 90 support, which the 840av has.

 

[ArmorAlley]

SCSI2SDv6 on internal 840av bus (bottom line)

vs

IBM 17gb 10k SCSI (Hitachi) on SEIV bus (top line)

FYI the SCSI2SDv6 is using a slower 6.2.5 firmware, as newer firmware cause corruption on write at the moment (is being fixed)

View attachment 30920

Would you run this again with the disks reversed, the 10K drive on the Q84av bus (assuming it supports 8-bit SCSI) and the SCSI2SDv6 on the 8-bit port of the SEIV please? This would give more info as to whether the limiting factor is the bus or the drive on the Q840av.

You may to put a system folder onto the 10K drive and bless it, in case you can't boot from SEIV.

 

[stormy]

ArmourAlley unfortunately I can't put that HDD on the 840av bus as it is a 68 pin LVD drive, the internal bus is 50 pin. The reason the SCSI2SDv6 is reporting such slow speed is because I am forced to use firmware 6.2.5 which is before they implemented large performance increases. Newer firmware's cause the SD card to corrupt on writes unfortunately, but they do double the speed. I am hoping the author might be able to fix this, he has been sharing beta firmwares with me.

 

[ArmorAlley]

ArmourAlley unfortunately I can't put that HDD on the 840av bus as it is a 68 pin LVD drive, the internal bus is 50 pin. The reason the SCSI2SDv6 is reporting such slow speed is because I am forced to use firmware 6.2.5 which is before they implemented large performance increases. Newer firmware's cause the SD card to corrupt on writes unfortunately, but they do double the speed. I am hoping the author might be able to fix this, he has been sharing beta firmwares with me.

Fair enough. I had assumed that you would have a 68-pin to 50-pin SCSI adapter on hand. They aren't that expensive (about $10 from China) and useful to have. There are also 80-pin to 50-pin adaptors that van be used on old SCA drives.

 

[Trash80toHP_Mini]

The reason the SCSI2SDv6 is reporting such slow speed is because I am forced to use firmware 6.2.5 which is before they implemented large performance increases. Newer firmware's cause the SD card to corrupt on writes unfortunately, but they do double the speed. I am hoping the author might be able to fix this, he has been sharing beta firmwares with me.

File this one under Stupid SCSI2SDv6 Tricks: While you're in touch with the developers, ask if this might be possible:

Hooking a pair of v6 units up to a (theoretical) controller board that would stripe their 8-bit Fast/Narrow data (in a RAID .5 configuration?) across the two SD Cards as 16-bit Fast/Wide to your SEIV? That would push sustained transfer rate by a factor of (nearly?) two, blowing the doors off the your 840AV's 5MB/s limited 8-bit SCSI Controller.

Ridiculous SCSI2SDv6 Trick would be utilizing SEIV's RAID capability to waste hundreds of dollars configuring several paired v6 units as RAID 0 to achieve the sustained transfer rate of your (now inexpensive) LVDS drive.

Stonkingly Stupid SCSI2SDv6 Trick would be spending several hundreds to thousands of dollars on such v6 unit RAID arrays to achieve the theoretical capability of the SEIV.

Coffee's kicking in, so morning madness abates. :blink: