VOGONS


First post, by RaiderOfLostVoodoo

User metadata
Rank Member
Rank
Member

So, I'm exchanging PMs with ChrisK about the ALive Dual.

Here's what he wrote:

By the way, I looked up what the problem with the SATA ports on the ASRock was.
The chipset's own ports only do SATA-I. That works, i.e. 130..140MB/s run through there, but they were not hotplug capable. For an eSata port (yes, there are people who use something like that 😁) so unfortunately unsuitable. The separate SATA-II ports are hotpluggable. The thing is, that except an old Samsung SSD470 SSD all newer SSDs (Samsung and Sandisk) run only in SATA-I mode. I don't know why. I have tried all possible driver versions and finally also different SATA cables, but always only SATA-I worked. Therefore the data rates are only ~140MB/s max.
The funny thing is that the Samsung SSD Magician even says that SATA-II is available. Only how to activate it remains a mystery.

Translated from Krautish to English by the awesome DeepL algorythm.

I'm currently testing my ALive Dual which I acquired recently and ran into the same problem.
My Samsung SSDs (830 + 840. Don't have any other SSDs.) also get detected as SATA-I and max out at 140MB/s. So it's not a defect on ChrisK's board.
Does anyone have an idea how we might solve this issue?

Reply 1 of 16, by chiveicrook

User metadata
Rank Newbie
Rank
Newbie

Older sata chipsets quite often have broken auto-negotiation, which results in newer sata3 drives falling back to sata1 or sometimes even not working at all. Unless disks support jumpering to sata2 (which is extremely rare, if there is a jumper it's usually to force sata1 for even older chipsets) nothing can be done. AFAIK ALive Dual use JMicron JMB363 (or some other derivative of JMB36x), which is notoriously finnicky and usually only works in sata2 with original sata2 drives.

Reply 3 of 16, by RaiderOfLostVoodoo

User metadata
Rank Member
Rank
Member
chiveicrook wrote on 2022-02-17, 08:32:

Older sata chipsets quite often have broken auto-negotiation, which results in newer sata3 drives falling back to sata1 or sometimes even not working at all. Unless disks support jumpering to sata2 (which is extremely rare, if there is a jumper it's usually to force sata1 for even older chipsets) nothing can be done. AFAIK ALive Dual use JMicron JMB363 (or some other derivative of JMB36x), which is notoriously finnicky and usually only works in sata2 with original sata2 drives.

Damn it. But thanks for the clarification.
Using a PCIe controller card isn't possible, because I can't use AGP and PCIe at the same time. And PCI is even slower than SATA1.
But I could use two 2TB SSDs (once they become affordable in a few years) in RAID on the SATA1 slots for games and a 256GB SATA2 SSD for Windows. 😁

Do you know any SSDs which can be jumpered to SATA2? Would be glad about any hints.

Reply 4 of 16, by chiveicrook

User metadata
Rank Newbie
Rank
Newbie
RaiderOfLostVoodoo wrote on 2022-02-17, 09:14:
Damn it. But thanks for the clarification. Using a PCIe controller card isn't possible, because I can't use AGP and PCIe at the […]
Show full quote
chiveicrook wrote on 2022-02-17, 08:32:

Older sata chipsets quite often have broken auto-negotiation, which results in newer sata3 drives falling back to sata1 or sometimes even not working at all. Unless disks support jumpering to sata2 (which is extremely rare, if there is a jumper it's usually to force sata1 for even older chipsets) nothing can be done. AFAIK ALive Dual use JMicron JMB363 (or some other derivative of JMB36x), which is notoriously finnicky and usually only works in sata2 with original sata2 drives.

Damn it. But thanks for the clarification.
Using a PCIe controller card isn't possible, because I can't use AGP and PCIe at the same time. And PCI is even slower than SATA1.
But I could use two 2TB SSDs (once they become affordable in a few years) in RAID on the SATA1 slots for games and a 256GB SATA2 SSD for Windows. 😁

Do you know any SSDs which can be jumpered to SATA2? Would be glad about any hints.

Unfortunately I don't know of any. By the time SSDs became more common SATA was already old and manufacturers didn't really care about broken implementations of early controllers. Even on HDDs sata2 jumpers were rare - I've only ever seen one HDD with sata2 jumper setting, most come with sata1 jumper.
Some HDDs allowed to limit speed via firmware or via manufacturer's tool but again, I don't know of any SSD capable of that.

I wouldn't really worry about it though - slightly faster OS drive + slightly slower data drives will be more than enough for a dated system. On the plus side you can buy cheaper, slower ssds 😀

Reply 5 of 16, by rasz_pl

User metadata
Rank l33t
Rank
l33t
RaiderOfLostVoodoo wrote on 2022-02-17, 09:14:
chiveicrook wrote on 2022-02-17, 08:32:

AFAIK ALive Dual use JMicron JMB363 (or some other derivative of JMB36x), which is notoriously finnicky and usually only works in sata2 with original sata2 drives.

Using a PCIe controller card isn't possible, because I can't use AGP and PCIe at the same time. And PCI is even slower than SATA1.
But I could use two 2TB SSDs (once they become affordable in a few years) in RAID on the SATA1 slots for games and a 256GB SATA2 SSD for Windows. 😁

JMB36x doesnt seem to support trim
but since its hanging off PCIE bus you could attempt to mod your motherboard like this https://hackaday.com/2020/07/01/adding-pcie-t … the-easier-way/

Open Source AT&T Globalyst/NCR/FIC 486-GAC-2 proprietary Cache Module reproduction

Reply 6 of 16, by ChrisK

User metadata
Rank Member
Rank
Member

Is TRIM dependent on the SATA-controller? Not a operating system thing?
Does anyone know of / have a PCIe-SATA card with this JMicron JMB363? Would like to know if this would be able to drive a common SSD with SATA-II speed. Then it could be a firmware thing within the mainboard BIOS.

Reply 7 of 16, by Hirsch

User metadata
Rank Newbie
Rank
Newbie

I googled a little bit about this topic and found:

A Samsung 470 Series 256 GB SSD can achieve 243 MB/s read and 233 MB/s write.
(Source: https://www.storagereview.com/review/samsung- … gb-mz5pa256hmdr)

A benchmark of the JMicron JMB363 onboard controller of an ASUS Rampage III with a Crucial RealSSD C300 64GB achieves 80 MB/s read and 75 MB/s write.
(Source (German): https://www.hardwareluxx.de/index.php/artikel … on.html?start=9)
(Link to benchmark result: https://www.hardwareluxx.de/images/stories/ga … perf_esata1.jpg)

The result looks rather poor, even for SATA-I. I ask myself whether a SSD connected to the SATA-I port of the nForce3-250 southbridge might be faster.
@ChrisK: Do you have benchmark results of the Samsung 470 SSD on the AliveDual-eSata2 for comparison?

If it turns out that the performance is better than the benchmark I found patching a new JMicron BIOS into the AliveDual-eSata2 might be worth trying. (I haven't done this before, but I'm not afraid since I have an external programmer.)

Reply 8 of 16, by Hirsch

User metadata
Rank Newbie
Rank
Newbie

Continuing googling this topic I found:

Mod-BIOS for JMicron JMB363 with TRIM enabled (German):
https://www.computerbase.de/forum/threads/inf … jmb363.1997792/

I found a cheap JMicron JMB363 card on Ebay (12.99 Euro) and just bought it in order to give it a try. I'll keep you up to date whether this works.

Reply 9 of 16, by RaiderOfLostVoodoo

User metadata
Rank Member
Rank
Member
kixs wrote on 2022-02-17, 08:38:

I wouldn't worry about it. SATA-1 is fast enough. Access time is what matters.

But Hirsch and I enjoy maxing out platforms.

Hirsch wrote on 2022-02-17, 21:14:

The result looks rather poor, even for SATA-I. I ask myself whether a SSD connected to the SATA-I port of the nForce3-250 southbridge might be faster.
@ChrisK: Do you have benchmark results of the Samsung 470 SSD on the AliveDual-eSata2 for comparison?

He did. 190MB/s he said.
Also noteworthy: The two SATA2 ports seem to share bandwith.
On the SATA1 ports I measured 130-140MB/s with CrystalDisk. That would be 260-280MB/s in RAID.

I had another idea:
- Samsung 470 256GB on 1st SATA2 port
- Hotswap on 2nd SATA2 port
- 2x 2TB SSD RAID on the SATA1 ports
- DVD drive on IDE
- 3 more 2TB drives on the IDE ports via adapter
- add a floppy drive for additional retro feels during boot
In the 2030s we will probably be able to purchase 2TBs SSDs for less than 50 bucks. The SSDs on the IDE ports could be used to swap games, which you don't intend to play in the near future. So you don't have to uninstall/reinstall. Just swap it back to the RAID array if necessary. 😁

Reply 10 of 16, by chiveicrook

User metadata
Rank Newbie
Rank
Newbie
RaiderOfLostVoodoo wrote on 2022-02-17, 23:18:

Also noteworthy: The two SATA2 ports seem to share bandwith.

The JMicron controller also hangs off PCIE 1.0 bus at 1x speed which limits it to around 2.5GT/s total which in turn translates to around 2Gbit/s of useful bandwidth.
It's more of a SATA "not quite 2" controller 😀

Reply 11 of 16, by ChrisK

User metadata
Rank Member
Rank
Member
chiveicrook wrote on 2022-02-18, 07:42:
RaiderOfLostVoodoo wrote on 2022-02-17, 23:18:

Also noteworthy: The two SATA2 ports seem to share bandwith.

The JMicron controller also hangs off PCIE 1.0 bus at 1x speed which limits it to around 2.5GT/s total which in turn translates to around 2Gbit/s of useful bandwidth.
It's more of a SATA "not quite 2" controller 😀

That's correct. The two SATA-ports of the JMB363 share one PCIe v1.0 x1 lane.
My old benchmarks say you can get around 190MB/s with a Samsung SSD470 128GB which runs at SATA-II. That's also the theoretical maximum for PCIe x1.
All newer SSDs I tested (Samsung 840/850/Sandisk X300s) only run at SATA-I with ~140MB/s max. That is also the maximum which can be achived when connecting the SSDs to the nForce SATA-ports.

I thought about what another native SATA-II SSD would behave like? I wouldn't wonder if it would also run at SATA-II just like the Samsung 470 (which is SATA-II).
Unfortunately I only have SATA-III SSDs besides the one 470.

RaiderOfLostVoodoo wrote on 2022-02-17, 23:18:

...
- 3 more 2TB drives on the IDE ports via adapter

Of course I have benchmarks for this 😉
With some SSD connected via a SATA to IDE adaptor (I have some JMicron one 🤫) you can reach 80-90MB/s on the IDE ports (one drive tested at a time).

Reply 12 of 16, by ChrisK

User metadata
Rank Member
Rank
Member
Hirsch wrote on 2022-02-17, 21:35:
Continuing googling this topic I found: […]
Show full quote

Continuing googling this topic I found:

Mod-BIOS for JMicron JMB363 with TRIM enabled (German):
https://www.computerbase.de/forum/threads/inf … jmb363.1997792/

I found a cheap JMicron JMB363 card on Ebay (12.99 Euro) and just bought it in order to give it a try. I'll keep you up to date whether this works.

The ALiveDual-eSATA2 already supports AHCI mode. Selectable in BIOS. I also feel to remember the Samung Magician tells that TRIM is active (have to check to be sure). So I'm not sure if a firmware upgrade would improve that much.
I'm also using the stock Windows drivers as they have shown to be the fastest. All JMicron driver versions I tested just decreased performance.

Reply 13 of 16, by RaiderOfLostVoodoo

User metadata
Rank Member
Rank
Member
ChrisK wrote on 2022-02-18, 13:38:

With some SSD connected via a SATA to IDE adaptor (I have some JMicron one 🤫) you can reach 80-90MB/s on the IDE ports (one drive tested at a time).

Shouldn't it be more, because of UDMA133?

Interesting:
The IDE controller supports RAID.
https://www.nvidia.com/en-us/drivers/nforce-ide-415-win2KXP/

This is gonna be one hell of a XP machine. 🤣

Reply 14 of 16, by janskjaer

User metadata
Rank Member
Rank
Member
RaiderOfLostVoodoo wrote on 2022-02-17, 07:09:

I'm currently testing my ALive Dual which I acquired recently and ran into the same problem.
My Samsung SSDs (830 + 840. Don't have any other SSDs.) also get detected as SATA-I and max out at 140MB/s. So it's not a defect on ChrisK's board.

By sheer coincidence, I had started to test the very same motherboard with a SATA-III SSD and found the same problem.
At first, I expected it to be either a faulty SSD or a setting that need to be changed on either the disk or in the BIOS.
Imagine my surprise when I found your post from just some weeks before.

RaiderOfLostVoodoo wrote on 2022-02-17, 09:14:

Using a PCIe controller card isn't possible, because I can't use AGP and PCIe at the same time. And PCI is even slower than SATA1.

Hmm, I didn't have this issue. My PCI-E slot has a USB 3.0 hub installed, and my AGP has either an GeForce 6200 or 7800GS installed. Both work fine simultaneously.

chiveicrook wrote on 2022-02-17, 12:14:

On the plus side you can buy cheaper, slower ssds 😀

I wish that were true. SATA-II SSDs sell for the same, if not more, than SATA-III SSDs.

Hirsch wrote on 2022-02-17, 21:35:

I found a cheap JMicron JMB363 card on Ebay (12.99 Euro) and just bought it in order to give it a try. I'll keep you up to date whether this works.

Did you manage to do this?

I can confirm this is a SATA-III SSD issue with the JMicron JMB363 controller, as I normally have 2 WD Green 2TB conventional disks on these ports and they reach maximum SATA-300 read/write speeds.

I've conducted the following tests with CrystalDiskMark. I ran 3 rounds for each category to determine an average measurement.
The results consist of a Sandisk X110 on both the nForce3 250 SATA-I and JMicron JMB363 SATA-II controllers (tested in both IDE and AHCI mode via the setting in BIOS).

Profile: Default
Test: 1 GiB (x5) [C: 32% (38/119GiB)]
Time: Measure 5 sec / Interval 5 sec
OS: Windows Vista Ultimate SP2 [6.0 Build 6002] (x64)

nForce3 250 SATA-I controller, tests 1-3:

[TEST_1]
[Read]
SEQ 1MiB (Q= 8, T= 1): 132.142 MB/s [ 126.0 IOPS] < 62823.28 us>
SEQ 1MiB (Q= 1, T= 1): 128.573 MB/s [ 122.6 IOPS] < 8136.05 us>
RND 4KiB (Q= 32, T= 1): 26.824 MB/s [ 6548.8 IOPS] < 4767.34 us>
RND 4KiB (Q= 1, T= 1): 20.085 MB/s [ 4903.6 IOPS] < 200.11 us>
[Write]
SEQ 1MiB (Q= 8, T= 1): 111.155 MB/s [ 106.0 IOPS] < 74589.45 us>
SEQ 1MiB (Q= 1, T= 1): 108.851 MB/s [ 103.8 IOPS] < 9609.20 us>
RND 4KiB (Q= 32, T= 1): 54.454 MB/s [ 13294.4 IOPS] < 2383.91 us>
RND 4KiB (Q= 1, T= 1): 31.723 MB/s [ 7744.9 IOPS] < 125.30 us>

[TEST_2]
[Read]
SEQ 1MiB (Q= 8, T= 1): 135.283 MB/s [ 129.0 IOPS] < 61252.23 us>
SEQ 1MiB (Q= 1, T= 1): 129.407 MB/s [ 123.4 IOPS] < 8082.32 us>
RND 4KiB (Q= 32, T= 1): 25.914 MB/s [ 6326.7 IOPS] < 5008.75 us>
RND 4KiB (Q= 1, T= 1): 17.445 MB/s [ 4259.0 IOPS] < 229.31 us>
[Write]
SEQ 1MiB (Q= 8, T= 1): 105.076 MB/s [ 100.2 IOPS] < 79289.40 us>
SEQ 1MiB (Q= 1, T= 1): 100.456 MB/s [ 95.8 IOPS] < 10419.12 us>
RND 4KiB (Q= 32, T= 1): 51.799 MB/s [ 12646.2 IOPS] < 2513.05 us>
RND 4KiB (Q= 1, T= 1): 25.251 MB/s [ 6164.8 IOPS] < 157.38 us>

[TEST_3]
[Read]
SEQ 1MiB (Q= 8, T= 1): 115.787 MB/s [ 110.4 IOPS] < 71823.99 us>
SEQ 1MiB (Q= 1, T= 1): 111.789 MB/s [ 106.6 IOPS] < 9365.28 us>
RND 4KiB (Q= 32, T= 1): 25.887 MB/s [ 6320.1 IOPS] < 5041.29 us>
RND 4KiB (Q= 1, T= 1): 16.904 MB/s [ 4127.0 IOPS] < 237.23 us>
[Write]
SEQ 1MiB (Q= 8, T= 1): 103.408 MB/s [ 98.6 IOPS] < 80084.44 us>
SEQ 1MiB (Q= 1, T= 1): 100.894 MB/s [ 96.2 IOPS] < 10371.72 us>
RND 4KiB (Q= 32, T= 1): 51.937 MB/s [ 12679.9 IOPS] < 2508.06 us>
RND 4KiB (Q= 1, T= 1): 25.044 MB/s [ 6114.3 IOPS] < 158.84 us>

JMicron JMB363 SATA-II controller (AHCI), tests 1-3:

[TEST_1]
[Read]
SEQ 1MiB (Q= 8, T= 1): 133.599 MB/s [ 127.4 IOPS] < 62366.91 us>
SEQ 1MiB (Q= 1, T= 1): 131.505 MB/s [ 125.4 IOPS] < 7966.70 us>
RND 4KiB (Q= 32, T= 1): 24.371 MB/s [ 5950.0 IOPS] < 5340.59 us>
RND 4KiB (Q= 1, T= 1): 18.443 MB/s [ 4502.7 IOPS] < 218.24 us>
[Write]
SEQ 1MiB (Q= 8, T= 1): 128.160 MB/s [ 122.2 IOPS] < 64779.25 us>
SEQ 1MiB (Q= 1, T= 1): 126.049 MB/s [ 120.2 IOPS] < 8298.51 us>
RND 4KiB (Q= 32, T= 1): 43.546 MB/s [ 10631.3 IOPS] < 2966.22 us>
RND 4KiB (Q= 1, T= 1): 27.683 MB/s [ 6758.5 IOPS] < 144.08 us>

[TEST_2]
[Read]
SEQ 1MiB (Q= 8, T= 1): 128.790 MB/s [ 122.8 IOPS] < 63850.65 us>
SEQ 1MiB (Q= 1, T= 1): 126.676 MB/s [ 120.8 IOPS] < 8263.50 us>
RND 4KiB (Q= 32, T= 1): 24.067 MB/s [ 5875.7 IOPS] < 5399.78 us>
RND 4KiB (Q= 1, T= 1): 16.316 MB/s [ 3983.4 IOPS] < 245.27 us>
[Write]
SEQ 1MiB (Q= 8, T= 1): 124.575 MB/s [ 118.8 IOPS] < 66422.98 us>
SEQ 1MiB (Q= 1, T= 1): 122.277 MB/s [ 116.6 IOPS] < 8549.56 us>
RND 4KiB (Q= 32, T= 1): 43.246 MB/s [ 10558.1 IOPS] < 3017.89 us>
RND 4KiB (Q= 1, T= 1): 26.730 MB/s [ 6525.9 IOPS] < 149.04 us>

[TEST_3]
[Read]
SEQ 1MiB (Q= 8, T= 1): 118.075 MB/s [ 112.6 IOPS] < 70387.68 us>
SEQ 1MiB (Q= 1, T= 1): 117.236 MB/s [ 111.8 IOPS] < 8822.94 us>
RND 4KiB (Q= 32, T= 1): 23.272 MB/s [ 5681.6 IOPS] < 5563.09 us>
RND 4KiB (Q= 1, T= 1): 16.107 MB/s [ 3932.4 IOPS] < 248.21 us>
[Write]
SEQ 1MiB (Q= 8, T= 1): 117.879 MB/s [ 112.4 IOPS] < 70434.69 us>
SEQ 1MiB (Q= 1, T= 1): 115.366 MB/s [ 110.0 IOPS] < 9066.10 us>
RND 4KiB (Q= 32, T= 1): 43.217 MB/s [ 10551.0 IOPS] < 3015.37 us>
RND 4KiB (Q= 1, T= 1): 25.357 MB/s [ 6190.7 IOPS] < 156.92 us>

JMicron JMB363 SATA-II controller (IDE), tests 1-3:

[TEST_1]
[Read]
SEQ 1MiB (Q= 8, T= 1): 118.923 MB/s [ 113.4 IOPS] < 69936.20 us>
SEQ 1MiB (Q= 1, T= 1): 116.415 MB/s [ 111.0 IOPS] < 8988.76 us>
RND 4KiB (Q= 32, T= 1): 24.452 MB/s [ 5969.7 IOPS] < 5343.95 us>
RND 4KiB (Q= 1, T= 1): 17.422 MB/s [ 4253.4 IOPS] < 231.16 us>
[Write]
SEQ 1MiB (Q= 8, T= 1): 121.445 MB/s [ 115.8 IOPS] < 68298.47 us>
SEQ 1MiB (Q= 1, T= 1): 119.353 MB/s [ 113.8 IOPS] < 8765.14 us>
RND 4KiB (Q= 32, T= 1): 42.916 MB/s [ 10477.5 IOPS] < 3001.62 us>
RND 4KiB (Q= 1, T= 1): 25.264 MB/s [ 6168.0 IOPS] < 158.21 us>
[TEST_2]
[Read]
SEQ 1MiB (Q= 8, T= 1): 131.089 MB/s [ 125.0 IOPS] < 63545.61 us>
SEQ 1MiB (Q= 1, T= 1): 123.960 MB/s [ 118.2 IOPS] < 8441.55 us>
RND 4KiB (Q= 32, T= 1): 24.538 MB/s [ 5990.7 IOPS] < 5281.74 us>
RND 4KiB (Q= 1, T= 1): 17.596 MB/s [ 4295.9 IOPS] < 228.87 us>
[Write]
SEQ 1MiB (Q= 8, T= 1): 121.640 MB/s [ 116.0 IOPS] < 68299.11 us>
SEQ 1MiB (Q= 1, T= 1): 119.337 MB/s [ 113.8 IOPS] < 8768.52 us>
RND 4KiB (Q= 32, T= 1): 43.722 MB/s [ 10674.3 IOPS] < 2966.23 us>
RND 4KiB (Q= 1, T= 1): 25.309 MB/s [ 6179.0 IOPS] < 158.00 us>
[TEST_3]
[Read]
SEQ 1MiB (Q= 8, T= 1): 119.143 MB/s [ 113.6 IOPS] < 69796.63 us>
SEQ 1MiB (Q= 1, T= 1): 116.619 MB/s [ 111.2 IOPS] < 8969.41 us>
RND 4KiB (Q= 32, T= 1): 24.494 MB/s [ 5980.0 IOPS] < 5212.59 us>
RND 4KiB (Q= 1, T= 1): 17.528 MB/s [ 4279.3 IOPS] < 229.79 us>
[Write]
SEQ 1MiB (Q= 8, T= 1): 121.652 MB/s [ 116.0 IOPS] < 68321.07 us>
SEQ 1MiB (Q= 1, T= 1): 119.557 MB/s [ 114.0 IOPS] < 8745.35 us>
RND 4KiB (Q= 32, T= 1): 43.026 MB/s [ 10504.4 IOPS] < 3003.32 us>
RND 4KiB (Q= 1, T= 1): 25.226 MB/s [ 6158.7 IOPS] < 158.46 us>

The AHCI and IDE tests should be no different as the setting in the BIOS is to allow disks to be hot-pluggable or not.
Next week, I will be able to test again with an Intel SSDSA2M160G2GC 160GB SATA II 3Gb/s. Hopefully, the results will be better than the Sandisk x110, and similar or better to those of the Samsung 470 series.

DELL Dimension XPS M200s
:Intel P1 MMX 200MHz
:64MB EDO
:DOS 6.22/Win95b
:Matrox Millenium II + m3D (PowerVR PCX2)
Chaintech 7VJL Apogee
:AMD AthlonXP 2700+
:512MB DDR
:Win98SE/2000 SP4
:3dfx Voodoo5 5500 AGP

Reply 15 of 16, by janskjaer

User metadata
Rank Member
Rank
Member

Intel SSDSA2M160G2GC 160GB

nForce3 250 SATA-I controller, tests 1-2:

[TEST_1]
[Read]
SEQ 1MiB (Q= 8, T= 1): 138.009 MB/s [ 131.6 IOPS] < 60284.64 us>
SEQ 1MiB (Q= 1, T= 1): 135.482 MB/s [ 129.2 IOPS] < 7726.93 us>
RND 4KiB (Q= 32, T= 1): 20.764 MB/s [ 5069.3 IOPS] < 6268.70 us>
RND 4KiB (Q= 1, T= 1): 16.537 MB/s [ 4037.4 IOPS] < 243.84 us>
[Write]
SEQ 1MiB (Q= 8, T= 1): 89.139 MB/s [ 85.0 IOPS] < 92867.87 us>
SEQ 1MiB (Q= 1, T= 1): 89.136 MB/s [ 85.0 IOPS] < 11744.07 us>
RND 4KiB (Q= 32, T= 1): 51.515 MB/s [ 12576.9 IOPS] < 2484.87 us>
RND 4KiB (Q= 1, T= 1): 31.491 MB/s [ 7688.2 IOPS] < 126.23 us>
[TEST_2]
[Read]
SEQ 1MiB (Q= 8, T= 1): 129.400 MB/s [ 123.4 IOPS] < 64342.42 us>
SEQ 1MiB (Q= 1, T= 1): 120.174 MB/s [ 114.6 IOPS] < 8713.09 us>
RND 4KiB (Q= 32, T= 1): 20.499 MB/s [ 5004.6 IOPS] < 6183.97 us>
RND 4KiB (Q= 1, T= 1): 15.548 MB/s [ 3795.9 IOPS] < 259.53 us>
[Write]
SEQ 1MiB (Q= 8, T= 1): 87.887 MB/s [ 83.8 IOPS] < 94135.59 us>
SEQ 1MiB (Q= 1, T= 1): 87.253 MB/s [ 83.2 IOPS] < 11991.34 us>
RND 4KiB (Q= 32, T= 1): 49.597 MB/s [ 12108.6 IOPS] < 2583.75 us>
RND 4KiB (Q= 1, T= 1): 28.037 MB/s [ 6845.0 IOPS] < 142.10 us>

SATA-I results overall are much slower than that of the Sandisk x100.

JMicron JMB363 SATA-II controller, Windows Vista SP2 driver, tests 1-2:

[TEST_1]
[Read]
SEQ 1MiB (Q= 8, T= 1): 177.229 MB/s [ 169.0 IOPS] < 42191.55 us>
SEQ 1MiB (Q= 1, T= 1): 166.952 MB/s [ 159.2 IOPS] < 6269.90 us>
RND 4KiB (Q= 32, T= 1): 12.713 MB/s [ 3103.8 IOPS] < 9961.85 us>
RND 4KiB (Q= 1, T= 1): 8.597 MB/s [ 2098.9 IOPS] < 472.51 us>
[Write]
SEQ 1MiB (Q= 8, T= 1): 99.210 MB/s [ 94.6 IOPS] < 83373.66 us>
SEQ 1MiB (Q= 1, T= 1): 97.538 MB/s [ 93.0 IOPS] < 10715.93 us>
RND 4KiB (Q= 32, T= 1): 12.812 MB/s [ 3127.9 IOPS] < 9883.43 us>
RND 4KiB (Q= 1, T= 1): 11.117 MB/s [ 2714.1 IOPS] < 364.09 us>
[TEST_2]
[Read]
SEQ 1MiB (Q= 8, T= 1): 177.416 MB/s [ 169.2 IOPS] < 46198.01 us>
SEQ 1MiB (Q= 1, T= 1): 170.301 MB/s [ 162.4 IOPS] < 6147.75 us>
RND 4KiB (Q= 32, T= 1): 12.696 MB/s [ 3099.6 IOPS] < 9975.47 us>
RND 4KiB (Q= 1, T= 1): 8.615 MB/s [ 2103.3 IOPS] < 471.50 us>
[Write]
SEQ 1MiB (Q= 8, T= 1): 99.414 MB/s [ 94.8 IOPS] < 83328.03 us>
SEQ 1MiB (Q= 1, T= 1): 97.533 MB/s [ 93.0 IOPS] < 10716.31 us>
RND 4KiB (Q= 32, T= 1): 12.867 MB/s [ 3141.4 IOPS] < 9883.01 us>
RND 4KiB (Q= 1, T= 1): 11.127 MB/s [ 2716.6 IOPS] < 364.23 us>

SATA-II speeds for sequential reads are immediately up, well past the limits reached by the Sandisk x110, peaking around 177Mb/s.
However, all write speeds and 4k reads are down in comparison to the Sandisk x110 connected to a SATA-I port.

JMicron JMB363 SATA-II controller, JMicron 1.17.65.11 driver, tests 1-2:

[TEST_1]
[Read]
SEQ 1MiB (Q= 8, T= 1): 183.927 MB/s [ 175.4 IOPS] < 45374.31 us>
SEQ 1MiB (Q= 1, T= 1): 178.285 MB/s [ 170.0 IOPS] < 5874.81 us>
RND 4KiB (Q= 32, T= 1): 23.119 MB/s [ 5644.3 IOPS] < 5482.67 us>
RND 4KiB (Q= 1, T= 1): 17.543 MB/s [ 4283.0 IOPS] < 229.47 us>
[Write]
SEQ 1MiB (Q= 8, T= 1): 96.688 MB/s [ 92.2 IOPS] < 85662.14 us>
SEQ 1MiB (Q= 1, T= 1): 96.687 MB/s [ 92.2 IOPS] < 10832.26 us>
RND 4KiB (Q= 32, T= 1): 46.971 MB/s [ 11467.5 IOPS] < 2723.64 us>
RND 4KiB (Q= 1, T= 1): 29.239 MB/s [ 7138.4 IOPS] < 136.08 us>
[TEST_2]
[Read]
SEQ 1MiB (Q= 8, T= 1): 183.748 MB/s [ 175.2 IOPS] < 45387.15 us>
SEQ 1MiB (Q= 1, T= 1): 178.685 MB/s [ 170.4 IOPS] < 5856.56 us>
RND 4KiB (Q= 32, T= 1): 20.144 MB/s [ 4918.0 IOPS] < 6369.70 us>
RND 4KiB (Q= 1, T= 1): 15.890 MB/s [ 3879.4 IOPS] < 253.77 us>
[Write]
SEQ 1MiB (Q= 8, T= 1): 97.320 MB/s [ 92.8 IOPS] < 84534.96 us>
SEQ 1MiB (Q= 1, T= 1): 97.535 MB/s [ 93.0 IOPS] < 10734.60 us>
RND 4KiB (Q= 32, T= 1): 47.040 MB/s [ 11484.4 IOPS] < 2745.99 us>
RND 4KiB (Q= 1, T= 1): 29.480 MB/s [ 7197.3 IOPS] < 135.04 us>

Contradictory to what @ChrisK found, the JMicron drivers performed much faster than the stock Windows drivers (Vista SP2).
Sequential read speeds are almost reaching the theoretical limit of 190Mb/s for this board.
4k writes are 3x faster than those on the Vista driver.

Not a bad performance increase for a SATA-II SSD that cost less than 10 EUR.

Based on the results, it appears that a native SATA-II drive will have performance increases.
However, given the age of such drives, some of the SATA-III drives may be competitive to native SATA-II drives, even at SATA-I speeds.

DELL Dimension XPS M200s
:Intel P1 MMX 200MHz
:64MB EDO
:DOS 6.22/Win95b
:Matrox Millenium II + m3D (PowerVR PCX2)
Chaintech 7VJL Apogee
:AMD AthlonXP 2700+
:512MB DDR
:Win98SE/2000 SP4
:3dfx Voodoo5 5500 AGP

Reply 16 of 16, by Hirsch

User metadata
Rank Newbie
Rank
Newbie
janskjaer wrote on 2022-03-03, 16:57:
Hirsch wrote on 2022-02-17, 21:35:

I found a cheap JMicron JMB363 card on Ebay (12.99 Euro) and just bought it in order to give it a try. I'll keep you up to date whether this works.

Did you manage to do this?

I managed to do the tests with the card from Ebay and a Samsung 830 Series SSD with 128 GB (SATA-III drive which can do 520 MB/s read and 320 MB/s write according to datasheet). I'll append three screenshots with:
- Stock ROM v1.03 which was on the card when it arrived.
- Last official ROM v1.07.24 from the JMicron FTP server (https://archive.org/details/ftp-driver.jmicron.com.tw).
- Mod-ROM v1.08.01 from https://www.computerbase.de/forum/threads/inf … jmb363.1997792/ (Do not use the first mod-ROM from JMB363 mod BIOS.zip, this one is broken!!! Rather take jm363new.zip)

I did 5 tests in order to get a good average. I noticed that the date displayed by the mod-ROM looks a little bit weird. It says "Copyright(C) 2005-2009" while the last official ROM with a lower revision number says "Copyright(C) 2005-2010".

Attachments