VOGONS


First post, by Babasha

User metadata
Rank Member
Rank
Member

Arrived without anything.

Made it DIY paddle-board for ISA slot, burn last firmware 2.09, install 4x SIMM’s, check and install.
Create every RAID, JBOD configuration supported by this controller with Seagate Medalist's 2.1Gb

IMG_2530.1640017820.jpgIMG_1612.1640017900.jpgIMG_1460.1640017931.jpgIMG_1521.1640017959.jpgIMG_1478.1640017986.jpgIMG_1515.1640018011.jpg

Need help? Begin with photo and model of your hardware! 😉

Reply 1 of 6, by Horun

User metadata
Rank l33t
Rank
l33t

Wow good job ! 🤣 it has the same processor as my 286 😀

Hate posting a reply and then have to edit it because it made no sense 😁 First computer was an IBM 3270 workstation with CGA monitor.

Reply 2 of 6, by Babasha

User metadata
Rank Member
Rank
Member
Babasha wrote on 2021-12-20, 16:34:
Arrived without anything. […]
Show full quote

Arrived without anything.

Made it DIY paddle-board for ISA slot, burn last firmware 2.09, install 4x SIMM’s, check and install.
Create every RAID, JBOD configuration supported by this controller with Seagate Medalist's 2.1Gb

IMG_2530.1640017820.jpgIMG_1612.1640017900.jpgIMG_1460.1640017931.jpgIMG_1521.1640017959.jpgIMG_1478.1640017986.jpgIMG_1515.1640018011.jpg

Just to remember - these controllers were produced in the first half and until the mid-90s, they are relevant (reveal themselves) on computers and disks of THAT time. No, don't expect a miracle if you have a Super Socket-7 / Slot-1 machine and a newer hard drive(s) than mid-late 1996. There is no need to chase the fastest IDE hard drives or solid-state drives by hanging them on such a controller (the whole point of these cards was to eliminate the weak points of the relatively slow IDE hard drives of THAT time).

Optimal configuration for PCI-variant Tekram DC-6X0C (D) - processor from 486DX2 to Pentium-1 133MHz, motherboard WITHOUT plug & play and highly desirable WITHOUT integrated IDE controller!
At the BIOS level, it is optimal for the motherboard to allow manually assigning interrupts and resources for each of the PCI slots (such were encountered in 1994-1995). Nevertheless, this controller can be "started" on less rare motherboards with plug & play, but the system will allocate resources at its discretion. Since PCI controllers Tekram DC-6X0C (D) are not quite plug & play (according to the authors themselves), it is fraught with possible conflicts with other PCI plug & play cards and devices in your computer. My advice is to minimize the number of boards, especially those that are "greedy" to the number of interrupts they use (such as USB/Firewire "two in one" multicontrollers that can select 3-4 interrupts for themselves and are guaranteed to capture 14 and or 15 interrupts, which will conflict with Tekram. If you have such an option and the board supports it, use an additional paddle-board (micro-board on a loop that is inserted into the "short" part of the ISA-16 connector) to correctly capture IRQ 14 and 15 interrupts for needs controller.

The controllers support the following RAID modes in hardware: "mirroring" RAID-1 and some analogue of JBOD (when the capacity of two hard disks is summed up into one logical disk). JBOD mode works only with pairs of disks on one of the IDE channels (Primary Master + Primary Slave or Secondary Master + Secondary Slave) of the controller. Mirroring works only with pairs of disks on different IDE channels. I did not find a drop in performance when using these modes.

Currently, the controller is used with 4MB of RAM and Seagate Medalist 2.1Gb hard drives in JBOD or "mirror" modes + single laptop Seagate 810Mb HDD. Configuration and testing were carried out under DOS (SS​​T 4.78, Hddspeed, AMIDiag) and partially under Windows 98 in ATTO Disk Bench 32. The hardest part is to interpret the test results. Apparently, these tests are strictly tailored for traditional controllers with hard drives and show incorrect results in some cases or do not carry out some of the tests at all. Everywhere in the tests, the performance of Cache Read Speed ​​is steadily growing from 5.8 - 6 Mb/s to 12 -12.4 Mb/s, and increase in Linear Write Speed ​​from 3.2 - 5.1 Mb/s to 5.7 - 6.1 Mb/s. Phenomenal drop in Liner Read Speed ​​from 3.6 - 5.3Mb / s up to 1.0 - 1.9Mb.

Personal conclusions.

Disk operations, during real work, do not feel as badly sagging performance and introduce lags as when using an integrated controller on the motherboard. It's nice to get a up to 16MB hardware Smartdrive that works both for reading and writing and does not consume the main RAM and processor. Hardware RAID-1 and JBOD - I don't even know what to compare with? I just don’t remember software products for DOS / Win9x with similar functionality, and if there were, the price for them is interesting and how many resources of the central processor of the P1-133MHz level did they “eat”? Plus, the controller itself has a internal cloning function - a kind of Norton Ghost or Partition Magic in cloning mode when setting up a numbers of computers (created an image of a working system and copy it onto a bunch of disks).

Yes, in the BIOS of the controller, you can change the minimum size of the cached block from 512 bytes (sector) to 8Kb (and this is already a file system cluster), which gives a performance increase on file systems, depending on the cluster size, up to 10-15% compared to the value for default.

Tests

The screenshots below show absolutely identical computer configuration (CPU INTEL P1-233, MB - VIA VP2, RAM - 48MB, VIDEO - S3 TRIO 3D 8MB PCI)

In the first case, testing was carried out with an integrated motherboard controller with max. parameters related to IDE
In the second case, testing was carried out with the PCI IDE CACHE controller TEKRAM DC-690CD with default settings (theoretically, you can still change the size of the cached block from 512 bytes to 8Kb, which gives a performance increase of up to 10-15% depending on the type of file system ).

And then the results of real-use performance: installation of Windows NT 4.0 on a Seagate ST32122A 2.1GB hard drive (installed on a FAT-16 partition with automatic conversion to NTFS
7 min 15 sec (on the integrated motherboard controller)
6 min 04 sec (on TEKRAM DC-690CD).

IMG_1633.1640253952.jpgIMG_1634.1640253993.jpg

Need help? Begin with photo and model of your hardware! 😉

Reply 3 of 6, by OMORES

User metadata
Rank Member
Rank
Member

Can you test this card on a newer system? I believe that Tekram DC-690CD has proper NT 3.x drivers. I'm curios if it works in a configuration where PCI slots are using a PCI-E to PCI bridge.

I'm looking for a PCI - IDE card with NT 3.1 drivers.

Reply 4 of 6, by Babasha

User metadata
Rank Member
Rank
Member
OMORES wrote on 2022-05-12, 15:07:

Can you test this card on a newer system? I believe that Tekram DC-690CD has proper NT 3.x drivers. I'm curios if it works in a configuration where PCI slots are using a PCI-E to PCI bridge.

I'm looking for a PCI - IDE card with NT 3.1 drivers.

Sorry, I don’t have any PC newer than P1-200MHz.

Need help? Begin with photo and model of your hardware! 😉

Reply 5 of 6, by Babasha

User metadata
Rank Member
Rank
Member
OMORES wrote on 2022-05-12, 15:07:

Can you test this card on a newer system? I believe that Tekram DC-690CD has proper NT 3.x drivers. I'm curios if it works in a configuration where PCI slots are using a PCI-E to PCI bridge.

I'm looking for a PCI - IDE card with NT 3.1 drivers.

One more. Tekram DC-6X0 is specific card and oriented on VERY early PCI realization, poorly compatible with anything higher than late 486/ early P1 motherboard and req’s ISA slot for paddle-board. So its bad idea to try it with modern (late 1996/7) motherboards.

Need help? Begin with photo and model of your hardware! 😉

Reply 6 of 6, by OMORES

User metadata
Rank Member
Rank
Member
Babasha wrote on 2022-05-12, 19:14:

One more. Tekram DC-6X0 is specific card and oriented on VERY early PCI realization, poorly compatible with anything higher than late 486/ early P1 motherboard and req’s ISA slot for paddle-board. So its bad idea to try it with modern (late 1996/7) motherboards.

Good to know, not on my list anymore.

Anyways, there must be some PCI - IDE adapter (around 1995-1996) - which is supported in NT 3.x but also works with NT 4 and 2000/XP.