I'd guess that the original board is the only external way to distinguish them. The original LSI board being only Pcie 2.0 capable, but the newer board with the extra componants or at least the pads for them, is PCIe 3 capable. However even with the 3.0 card it's still a crapshoot whether it comes with the PCIe 3 chipset.
Yeah, that's the thing... the previous video I showed a HP H221, labeled as 9207-8e, has the pads for those "extra components", but has the Rev B0 PCIe 2.0 only chipset. The H221 in this video, visually identical in every way, but has the Rev D1 PCIe 3.0 chipset.
Thanks for the info. HP is not the only one where things are not always what they are marked. parts for industrial robots and PLC's often have this issue. I trusted the card I got with a used Supermicro system was what was advertised without really verifying beyond visually looking at it. Used the first 2 lspci commands and it is what was advertised, so that's good.
I've only seen this confusion with HP cards, and specifically this H221 card. Other LSI cards usually have distinguished model numbers (9201-8i, 9207-8i, etc.) that are consistent. These HP H221 9207-8e apparently can be like a LSI 9207-8e, or a LSI 9205-8e.
One thing to check is if it really matters, does the PCI-E 3.0 really work on these cards. Reason for me saying this is that there were instances back then of early 3.0 cards, all reporting and working on 3.0, but they never got the 3.0 speeds, even when you started to lane restrict them. (Tape off PCI-E lanes)
I don't have any experience with that problem. I do know that the PCIe 3.0 LSI SAS2308 based HBAs can exceed 32Gbps (PCIe 2.0 x8), so the PCIe 3.0 x8 does make a difference. I have not seen any setup with a single HBA running at PCIe 3.0 x8 get close to 60Gbps though...
Weird timing for this video, a few weeks ago I was trying to flash a couple of H221 to the newest firmware trying to get them to play nice with an MD1200. I didn't get them to be friends but I did eventually get one to flash. After putting several days into research, chasing down files, and foolishly bricking one of them because of some dangerous instructions (step 1, delete everything; I should have thought twice), I've come to the conclusion that these are just strange cards. They've got a lot of odd corporate maneuvering in them, information is often conflicting because there are these significant variations, but there's also a precursor card I ran into some text about, which uses the same chips but then LSI renumbered it and HP used both with different features enabled on the precursor, the H221, and the H222? One of my cards identified itself as an H222 in the control interface and sas2flash
If your H221 has a Rev B0 chipset, you'll need to use the firmware from the 9205-8e. If it has a Rev D1 chipset, then you'll need to use the firmware from the 9207-8e. That's what is so strange about this HP H221 card that it has no consistency. As far as I know the HP H222 is a 9217-4i4e isn't it?
@@ArtofServerAll I know is I killed a card labelled 9207-8e, which identified itself in software as an H222, and I'm really unclear what socket layouts an H222 can have but most of the images I can find don't have 2 external sockets. I'll have to check out that firmware detail, that's not something I found before, nobody else seems to be talking about board Revisions. I got the 9207-8e firmware onto the surviving card using sas2flash P14 DOS, nothing else was willing to, and maybe they were right. I don't know what the problem is but I'm getting consistent controller errors, like several times per minute, and every few minutes the MD1200 will reset, setting the fans back to default extremely loud speeds
i'm a bit confused about the it mode controllers , what does that mean ? i'm guessing you wouldn't want to do that with an HP or DELL server is it a similar thing with other components like SSDs/HDDs or other PCIe cards/controllers ?
IT mode stands for Initiator Target mode and is all related to old school SCSI terms and way of functioning. I believe Art of Server has a video that covers it in more detail, however it basically means it acts just like a normal SATA controller on the motherboard does, it presents disks directly. RAID cards on the other hand may or may not pending on how they are configured by you the user.
No they're all old b0 but they work really well. Also got an lsi 9207-8e controlling 48 (now 24) drives with hba for a school nas (we actually don't have so much the budget
9205-8e would be PCIe 2.0, 9207-8e is PCIe 3.0 ... according to Broadcom specification. EDIT: according to chatGPT 9207-8e was introduced as PCIe 3.0, and there was never PCIe 2.0 version of it. I know it is possible to fake most of those details reported by SAS controller. Maybe HP is advertizing other SAS HBA as 9207-8e for compatibility reasons with the OS.
I would never rely on ChatGPT for anything that requires a bit more sophistication than a 1st grader to understand. Here's a summary: LSI SAS2308 Rev B0 chipset: Used in 9205-8e, supports only PCIe 2.0 x8 LSI SAS2308 Rev D1 chipset: Used in 9207-8e (LSI, non-HP version), supports PCIe 3.0 x8 HP H221, labeled 9207-8e, apparently can have SAS2308 Rev B0, and sometimes even SAS2308 Rev D1.
According to somebody on the internet that I can't point at anymore, 9205-8e had PCIe 3 capability, but it was an early release so LSI wasn't sure it worked or that the firmware was good for it; when they were more confident they renamed it to 9207-8e and HP just sometimes still put 9205-8e chips on H221 boards for a while. As far as I know the only distinction is what the firmware allows and the PCB design supports. The firmware determines how the chipset identifies itself in software, so you'd have to remove the heatsink and identify the actual chip to know.
@@Ziraya0 I don't think that the firmware can influence chip stepping (whether D1 or B0) though how it IDs the chip can be wonky at times. There's a blog with a guy whose H221 showed SAS2308_2(D1) on HP firmware but then showed SAS2308_1(D1) on stock LSI firmware. Note the minor version changed from 2 to 1. He took off the heatsink and the chip *was* indeed a -1 part so it's likely down to a bug in the HP firmware.
Well, PCIe 2.0 x8 is limited to about 32Gbps, while PCIe 3.0 x8 is limited to about 62Gbps. So, if your aggregate bandwidth consumption exceeds 32Gbps, then yes there's a real performance difference. This won't matter in most HDD use cases, but for an array of SSDs, it's easy to exceed that limit.
I'd guess that the original board is the only external way to distinguish them. The original LSI board being only Pcie 2.0 capable, but the newer board with the extra componants or at least the pads for them, is PCIe 3 capable. However even with the 3.0 card it's still a crapshoot whether it comes with the PCIe 3 chipset.
Yeah, that's the thing... the previous video I showed a HP H221, labeled as 9207-8e, has the pads for those "extra components", but has the Rev B0 PCIe 2.0 only chipset. The H221 in this video, visually identical in every way, but has the Rev D1 PCIe 3.0 chipset.
Thanks for the info. HP is not the only one where things are not always what they are marked. parts for industrial robots and PLC's often have this issue. I trusted the card I got with a used Supermicro system was what was advertised without really verifying beyond visually looking at it. Used the first 2 lspci commands and it is what was advertised, so that's good.
I've only seen this confusion with HP cards, and specifically this H221 card. Other LSI cards usually have distinguished model numbers (9201-8i, 9207-8i, etc.) that are consistent. These HP H221 9207-8e apparently can be like a LSI 9207-8e, or a LSI 9205-8e.
One thing to check is if it really matters, does the PCI-E 3.0 really work on these cards.
Reason for me saying this is that there were instances back then of early 3.0 cards, all reporting and working on 3.0, but they never got the 3.0 speeds, even when you started to lane restrict them. (Tape off PCI-E lanes)
I don't have any experience with that problem. I do know that the PCIe 3.0 LSI SAS2308 based HBAs can exceed 32Gbps (PCIe 2.0 x8), so the PCIe 3.0 x8 does make a difference. I have not seen any setup with a single HBA running at PCIe 3.0 x8 get close to 60Gbps though...
i got two of those H221 B0. i replaced them now by two avago lsi 9207-8e
nice!
Weird timing for this video, a few weeks ago I was trying to flash a couple of H221 to the newest firmware trying to get them to play nice with an MD1200. I didn't get them to be friends but I did eventually get one to flash. After putting several days into research, chasing down files, and foolishly bricking one of them because of some dangerous instructions (step 1, delete everything; I should have thought twice), I've come to the conclusion that these are just strange cards. They've got a lot of odd corporate maneuvering in them, information is often conflicting because there are these significant variations, but there's also a precursor card I ran into some text about, which uses the same chips but then LSI renumbered it and HP used both with different features enabled on the precursor, the H221, and the H222? One of my cards identified itself as an H222 in the control interface and sas2flash
If your H221 has a Rev B0 chipset, you'll need to use the firmware from the 9205-8e. If it has a Rev D1 chipset, then you'll need to use the firmware from the 9207-8e. That's what is so strange about this HP H221 card that it has no consistency.
As far as I know the HP H222 is a 9217-4i4e isn't it?
@@ArtofServerAll I know is I killed a card labelled 9207-8e, which identified itself in software as an H222, and I'm really unclear what socket layouts an H222 can have but most of the images I can find don't have 2 external sockets.
I'll have to check out that firmware detail, that's not something I found before, nobody else seems to be talking about board Revisions.
I got the 9207-8e firmware onto the surviving card using sas2flash P14 DOS, nothing else was willing to, and maybe they were right. I don't know what the problem is but I'm getting consistent controller errors, like several times per minute, and every few minutes the MD1200 will reset, setting the fans back to default extremely loud speeds
Next video is going to be about network cards with the same MAC
Lol. I don't think I've ever seen that before! Have you run into that?
I have actually got some- Realtek had some problems with that. It was very strange back then. :-)
@@ArtofServer it famously happened in the late 90s with the ne2000 clones. a bunch had the same MAC.
In your ebay listing, you specify PCI 3, does that mean the listing is for a Rev D1 controller card?
Yes, I'm very specific about what I sell in my store. You can always count on my listing to be specific about the details.
i'm a bit confused about the it mode controllers , what does that mean ? i'm guessing you wouldn't want to do that with an HP or DELL server
is it a similar thing with other components like SSDs/HDDs or other PCIe cards/controllers ?
IT mode stands for Initiator Target mode and is all related to old school SCSI terms and way of functioning.
I believe Art of Server has a video that covers it in more detail, however it basically means it acts just like a normal SATA controller on the motherboard does, it presents disks directly.
RAID cards on the other hand may or may not pending on how they are configured by you the user.
This video might help explain: th-cam.com/video/xEbQohy6v8U/w-d-xo.html
I have 3 of these!
The HP H221 with Rev D1 chipset?
No they're all old b0 but they work really well. Also got an lsi 9207-8e controlling 48 (now 24) drives with hba for a school nas (we actually don't have so much the budget
9205-8e would be PCIe 2.0, 9207-8e is PCIe 3.0 ... according to Broadcom specification.
EDIT: according to chatGPT 9207-8e was introduced as PCIe 3.0, and there was never PCIe 2.0 version of it.
I know it is possible to fake most of those details reported by SAS controller. Maybe HP is advertizing other SAS HBA as 9207-8e for compatibility reasons with the OS.
I would never rely on ChatGPT for anything that requires a bit more sophistication than a 1st grader to understand.
Here's a summary:
LSI SAS2308 Rev B0 chipset: Used in 9205-8e, supports only PCIe 2.0 x8
LSI SAS2308 Rev D1 chipset: Used in 9207-8e (LSI, non-HP version), supports PCIe 3.0 x8
HP H221, labeled 9207-8e, apparently can have SAS2308 Rev B0, and sometimes even SAS2308 Rev D1.
@@ArtofServer broadcom documents kind of confirm that. HP seems to be doing some shady business with those HBAs.
According to somebody on the internet that I can't point at anymore, 9205-8e had PCIe 3 capability, but it was an early release so LSI wasn't sure it worked or that the firmware was good for it; when they were more confident they renamed it to 9207-8e and HP just sometimes still put 9205-8e chips on H221 boards for a while. As far as I know the only distinction is what the firmware allows and the PCB design supports. The firmware determines how the chipset identifies itself in software, so you'd have to remove the heatsink and identify the actual chip to know.
@@Ziraya0 I don't think that the firmware can influence chip stepping (whether D1 or B0) though how it IDs the chip can be wonky at times. There's a blog with a guy whose H221 showed SAS2308_2(D1) on HP firmware but then showed SAS2308_1(D1) on stock LSI firmware. Note the minor version changed from 2 to 1. He took off the heatsink and the chip *was* indeed a -1 part so it's likely down to a bug in the HP firmware.
Are there real performance differences?
Well, PCIe 2.0 x8 is limited to about 32Gbps, while PCIe 3.0 x8 is limited to about 62Gbps. So, if your aggregate bandwidth consumption exceeds 32Gbps, then yes there's a real performance difference. This won't matter in most HDD use cases, but for an array of SSDs, it's easy to exceed that limit.
Now that pcie 5.0 is a thing.
PCIe 4.0 seemed really short lived, don't you think?