Gigabyte removes PCIe 5.0 support from B650 motherboards in latest BIOS update(videocardz.com)
videocardz.com
Gigabyte removes PCIe 5.0 support from B650 motherboards in latest BIOS update
https://videocardz.com/newz/gigabyte-removes-unofficial-pcie-5-0-support-from-b650-motherboards-in-latest-bios-update
50 comments
This happened on AM4 as well. https://www.techpowerup.com/258044/amds-latest-agesa-update-...
It should be illegal to retroactively take back features from products that you've already sold, regardless of whether the feature was advertised or how buggy it is.
The HN title does not match the actual article title. Didn't look super closely, but I don't see anything in the article suggesting the article title was changed after publishing.
> Gigabyte removes unofficial PCIe 5.0 support from B650 motherboards in latest BIOS update
It seems that the board was actually running in PCIe 5.0 mode, even though it was only supposed to (by the specs/manual) do PCIe 4.0
It was never advertised as supporting 5.0
> Gigabyte removes unofficial PCIe 5.0 support from B650 motherboards in latest BIOS update
It seems that the board was actually running in PCIe 5.0 mode, even though it was only supposed to (by the specs/manual) do PCIe 4.0
It was never advertised as supporting 5.0
Right, only the B650E chipset has official PCIe 5.0 support. The B650 may support it, but not officially.
This was the root of the article in my opinion:
> The issue may have emerged from the fact that PCIe 5.0 was enabled by default, and not clearly marked as experimental, and that may be something AMD did not like to see.
However, it seems to just be speculation. I too would like to know what the real reason was. Was Gigabyte dealing with returns/RMAs due to the feature? Was there pressure from AMD to stop allowing their lesser chipset to access PCIe 5.0?
This was the root of the article in my opinion:
> The issue may have emerged from the fact that PCIe 5.0 was enabled by default, and not clearly marked as experimental, and that may be something AMD did not like to see.
However, it seems to just be speculation. I too would like to know what the real reason was. Was Gigabyte dealing with returns/RMAs due to the feature? Was there pressure from AMD to stop allowing their lesser chipset to access PCIe 5.0?
I remember seeing a lot of people reporting instability & crashes that were tracked down to PCIe 5.0 being enabled. The signal integrity requirements are higher with PCIe5, so if the motherboard isn't actually specced and verified to support it in the factory, then i imagine it's kind of a lottery, where some pass but many crash and it will just cause headaches for users.
> The B650 may support it, but not officially.
My guess (having some high-speed design and manufacturing experience) is that the boards were physically designed and qualified around the official spec, which was PCIe 4.0.
Then after some confusion they discovered that the BIOS team had enabled PCIe 5.0 on boards that were never qualified for it. Batch-to-batch variations of the boards could have caused instability because the manufacturing materials and test processes were only targeting PCIe 4.0
My guess (having some high-speed design and manufacturing experience) is that the boards were physically designed and qualified around the official spec, which was PCIe 4.0.
Then after some confusion they discovered that the BIOS team had enabled PCIe 5.0 on boards that were never qualified for it. Batch-to-batch variations of the boards could have caused instability because the manufacturing materials and test processes were only targeting PCIe 4.0
> Was there pressure from AMD to stop allowing their lesser chipset to access PCIe 5.0?
Would be my guess. The fact that the B650 can even do 5.0 on the GPU lanes (it is able to do 4x 5.0 on NVME) makes me wonder if the B650 chipset is the same silicon as B650E and the speed is just really a configuration setting. Across the board the B650 seems like it is the exact same as a de-rated B650E.
It would surprise me if motherboard vendors who support 4.0 weren't actually compliant with 5.0 from a signaling / integrity perspective for future proofing and increased headroom.
Would be my guess. The fact that the B650 can even do 5.0 on the GPU lanes (it is able to do 4x 5.0 on NVME) makes me wonder if the B650 chipset is the same silicon as B650E and the speed is just really a configuration setting. Across the board the B650 seems like it is the exact same as a de-rated B650E.
It would surprise me if motherboard vendors who support 4.0 weren't actually compliant with 5.0 from a signaling / integrity perspective for future proofing and increased headroom.
> The fact that the B650 can even do 5.0 on the GPU lanes (it is able to do 4x 5.0 on NVME) makes me wonder if the B650 chipset is the same silicon as B650E and the speed is just really a configuration setting.
All the AM5 chipsets are the same silicon, except A620 which has less capability (might just be fused off though) and the B840 which seems to be a rebranded AM4 chipset. B650, B650E, B850, and X870 are all one Promontory 21 chipset; X670, X670E and X870E are two Promonotory 21 chipsets (one chained behind the other). The B840 seems to be the Promontory 19 silicon, same as a B550 AM4 chipset; which seems weird, but it has more downstream connectivity than the A620 chipset, although the cpu to chipset link runs at pci-e 4.0 for A620 and 3.0 for B550 and presumably B840, so that's a bottleneck if you have high throughput devices behind the chipset.
The GPU lanes on AM5 all come direct from the CPU, as well as at least one x4 m.2 socket, and there's 4 more CPU lanes not earmarked by AMD. Those don't touch the chipset, and any correlation between the chipset branding and the speed of those lanes is a marketing restriction: If you put B650E branding on your board, AMD says the board must support pcie 5.0 for the GPU and first NVMe slot; but the chipset isn't involved in those lanes, it's just a marketing agreement; I'm not sure if AMD disallows using pcie 5.0 on a B650 board for the GPU slot, but capability is a matter of trace design between the slots and the cpu and maybe what the board maker is willing to certify / what the firmware is willing to enable. For B850, AMD says pci-e 4.0 for the GPU and 5.0 for the first nvme.
I'm sure there are some X870 boards that are significantly different from B850 boards, but there's probably a lot of boards where everything is the same other than the product name. Likely the same was true of B650 and B650E. For reliable PCI-e 5.0 operation, you do need to follow proper design rules for the traces, and that results in a more expensive board than following the design rules for PCI-e 4.0 operation. But it can make product sense to sell some of those 5.0 capable boards with a lesser chipset brand anyway.
All the AM5 chipsets are the same silicon, except A620 which has less capability (might just be fused off though) and the B840 which seems to be a rebranded AM4 chipset. B650, B650E, B850, and X870 are all one Promontory 21 chipset; X670, X670E and X870E are two Promonotory 21 chipsets (one chained behind the other). The B840 seems to be the Promontory 19 silicon, same as a B550 AM4 chipset; which seems weird, but it has more downstream connectivity than the A620 chipset, although the cpu to chipset link runs at pci-e 4.0 for A620 and 3.0 for B550 and presumably B840, so that's a bottleneck if you have high throughput devices behind the chipset.
The GPU lanes on AM5 all come direct from the CPU, as well as at least one x4 m.2 socket, and there's 4 more CPU lanes not earmarked by AMD. Those don't touch the chipset, and any correlation between the chipset branding and the speed of those lanes is a marketing restriction: If you put B650E branding on your board, AMD says the board must support pcie 5.0 for the GPU and first NVMe slot; but the chipset isn't involved in those lanes, it's just a marketing agreement; I'm not sure if AMD disallows using pcie 5.0 on a B650 board for the GPU slot, but capability is a matter of trace design between the slots and the cpu and maybe what the board maker is willing to certify / what the firmware is willing to enable. For B850, AMD says pci-e 4.0 for the GPU and 5.0 for the first nvme.
I'm sure there are some X870 boards that are significantly different from B850 boards, but there's probably a lot of boards where everything is the same other than the product name. Likely the same was true of B650 and B650E. For reliable PCI-e 5.0 operation, you do need to follow proper design rules for the traces, and that results in a more expensive board than following the design rules for PCI-e 4.0 operation. But it can make product sense to sell some of those 5.0 capable boards with a lesser chipset brand anyway.
The physical motherboard PCB needs to support PCIe 5.0 speeds and be tested for it at the factory.
If their hardware team designed the PCB and the qualification process to the specifications of the B650 chipset, they would have targeted PCIe 4.0 speeds.
If their hardware team designed the PCB and the qualification process to the specifications of the B650 chipset, they would have targeted PCIe 4.0 speeds.
There's no need to test for it at the factory if you're not promising it works. Certainly, the PCB needs to be built for it; but I'm sure some boards are designed for B650E and then put together with a B650.
Regardless, the chipset has been out for almost 3 years. Kind of late to put the horse back in the barn. Regardless of certification, lots of systems are out there where it's enabled, and those with those systems and a pci-e 5.0 can tell us how it works regardless of what the branding says. Maybe it's flaky, maybe it rarely works. Maybe it just made sense to follow the pci-e 5.0 design rules for the gpu slots, even without the branding; maybe the pci-e 4.0 design rules are good enough at the trace lengths on a typical am5 board. Maybe once you have the PCB designed and built to run the cpu m.2 at 5.0, it doesn't cost anything to follow design rules for the cpu x16 slot as well.
Regardless, the chipset has been out for almost 3 years. Kind of late to put the horse back in the barn. Regardless of certification, lots of systems are out there where it's enabled, and those with those systems and a pci-e 5.0 can tell us how it works regardless of what the branding says. Maybe it's flaky, maybe it rarely works. Maybe it just made sense to follow the pci-e 5.0 design rules for the gpu slots, even without the branding; maybe the pci-e 4.0 design rules are good enough at the trace lengths on a typical am5 board. Maybe once you have the PCB designed and built to run the cpu m.2 at 5.0, it doesn't cost anything to follow design rules for the cpu x16 slot as well.
>B650 can even do 5.0
No AM5 chipset can do 5.0 speeds, they are all linked via x4 PCIe 4.0 lanes. What you're thinking about is PCIe 5.0 speeds from CPU to SSDs/GPU. There's no chipset that does PCIe 5.0 speeds, they are all on 4.0.
No AM5 chipset can do 5.0 speeds, they are all linked via x4 PCIe 4.0 lanes. What you're thinking about is PCIe 5.0 speeds from CPU to SSDs/GPU. There's no chipset that does PCIe 5.0 speeds, they are all on 4.0.
I would be surprised if they fully supported 5.0, because it's likely more expensive to maintain the 5.0 performance, and even $0.01 per mobo shipped could be millions of dollars in lost profit.
> and even $0.01 per mobo shipped could be millions of dollars in lost profit.
Gigabyte shipped 4.8 million MBs in the first quarter.
Assuming they sell 20 million MBs in a year, how does a $0.01 additional component cause $1 million in losses? By my math, it's 200k per year.
Further, what restricts gigabyte from raising the price of their MB by $0.10? or even $1 to accommodate the various $0.01 components they might need per board? Do you think the market wouldn't bear such a price variance for $200 boards?
Gigabyte shipped 4.8 million MBs in the first quarter.
Assuming they sell 20 million MBs in a year, how does a $0.01 additional component cause $1 million in losses? By my math, it's 200k per year.
Further, what restricts gigabyte from raising the price of their MB by $0.10? or even $1 to accommodate the various $0.01 components they might need per board? Do you think the market wouldn't bear such a price variance for $200 boards?
The B650 and B650E southbridges are probably the same but that doesn't mean the boards are the same. PCIe 5.0 has tighter tolerances which requires more expensive PCBs. Motherboard vendors don't future-proof low-end products when they can upsell it to you instead.
The board already supports 5.0, and the stack up requirements for PCIE 4.0/5.0 haven't notably changed. So, I don't think this is a PCB cost question, more of a design time question.
>Right, only the B650E chipset has official PCIe 5.0 support.
There is no AM5 chipset with PCIe 5.0 support. They are all on PCIe 4.0. The lanes come from CPU, they go to SSDs/GPU/chipset. All AM5 chipsets are linked to CPU via PCIe 4.0 lanes, none support PCIe 5.0 speeds.
edit: Not sure why the downvote, this is common information and it should be easy to link an AM5 motherboard that has x4 PCIe 5.0 downlink lanes to chipset. But they do not exist.
There is no AM5 chipset with PCIe 5.0 support. They are all on PCIe 4.0. The lanes come from CPU, they go to SSDs/GPU/chipset. All AM5 chipsets are linked to CPU via PCIe 4.0 lanes, none support PCIe 5.0 speeds.
edit: Not sure why the downvote, this is common information and it should be easy to link an AM5 motherboard that has x4 PCIe 5.0 downlink lanes to chipset. But they do not exist.
I've flagged for the misleading title. Removing "unofficial" was quite unjust.
We absolutely want to hold companies accountable, but if we lambast them for reasonable actions, where's the motivation for them to do right?
We absolutely want to hold companies accountable, but if we lambast them for reasonable actions, where's the motivation for them to do right?
At least users can go back to older firmware and get 5.0 back.
There was a story not long ago about some exercise equipment (I think) that remotely bricked the devices (when the company was being shutdown or something) though they could have worked offline just fine without that final update.
There was a story not long ago about some exercise equipment (I think) that remotely bricked the devices (when the company was being shutdown or something) though they could have worked offline just fine without that final update.
Lately, some BIOS updates have been advertised as not revertable. I'm sure you can still program the old firmware with an external programmer, but "for security reasons", the in-system update won't let you revert some updates that fix security issues.
Even without that barrier, it's not great when you have to pick between things like the bus speed and memory stability or other updates you want.
Even without that barrier, it's not great when you have to pick between things like the bus speed and memory stability or other updates you want.
I almost never install BIOS updates unless I'm encountering a specific issue. Vendors tend to discourage it as well. It's just one of those things where "if it works, don't fuck with it" is good advice.
IMHO, anytime they write "stability" or "memory compatibility" is a pretty compelling reason to update. Modern systems run pretty close to the edge, and DDR5 has some nasty training times that maybe an update could help with. How much you care about security fixes for microcode loading or boot time stuff is up to you... some of that is pretty esoteric, but some people are compelled to have all known security fixes.
Updates to support new CPUs are a maybe... I can probably wait and run that update right before swapping cpus, but some people might prefer to run them now, in case their cpu fails and they need to replace it ... a lot of boards can flash without a cpu installed, but you get a lot less feedback, it's nicer to flash when things are working.
Updates to support new CPUs are a maybe... I can probably wait and run that update right before swapping cpus, but some people might prefer to run them now, in case their cpu fails and they need to replace it ... a lot of boards can flash without a cpu installed, but you get a lot less feedback, it's nicer to flash when things are working.
> DDR5 has some nasty training times
I was having a hard time figuring out what’s wrong with my new system, until I read an offhand comment about DDR5 having long training times on AMD, and I should just leave the system running for 5-10 minutes… sounded like an urban legend, but turned out to be reality.
I was having a hard time figuring out what’s wrong with my new system, until I read an offhand comment about DDR5 having long training times on AMD, and I should just leave the system running for 5-10 minutes… sounded like an urban legend, but turned out to be reality.
Yeah, it's super frustrating, I'm happy I knew about it before I did my ddr5 build. And then my board has this cool feature where when I load the XMP/EXPO profile from the ram, it turns termination from Auto to Off, and won't boot... sometimes it figures it out and goes back to the JEDEC profile, but usually I have to clear the CMOS. Tons of fun! And latest BIOS does the same thing, I updated recently cause I was having weird things, which did go away but not sure if the BIOS update or new fan settings helped more.
I keep updating BIOS because newer versions often provide stability improvements and let you push higher overclocks. That in itself is not very interesting for an everyday overclock (which shouldn't push the hardware limits), but do allow for lowering the voltages used for sensitive components (like VSoC which was responsible for burning CPUs a couple of years ago). Lower voltage means safer operation and slower silicon degradation.
For example, I'm now running 6000 MT/s on stock VSoC which definitely wasn't possible on my hardware 1.5 years ago.
That said, my mobo vendor of choice is not known for doing rug pulls (unlike Gigabyte, who are also fond of releasing several board revisions under the same model with significant differences between them). It's the opposite — for example, a recent BIOS update added support for running internal graphics at 4K/120Hz on the same hardware that could only do 4K/60Hz before.
For example, I'm now running 6000 MT/s on stock VSoC which definitely wasn't possible on my hardware 1.5 years ago.
That said, my mobo vendor of choice is not known for doing rug pulls (unlike Gigabyte, who are also fond of releasing several board revisions under the same model with significant differences between them). It's the opposite — for example, a recent BIOS update added support for running internal graphics at 4K/120Hz on the same hardware that could only do 4K/60Hz before.
Makes sense if you're into that I guess. I never change the BIOS settings from their defaults unless something is causing a problem. I definitely don't fool with overclocking.
You still might want to look into curve optimizer, which is just a marketing term for an automatic undervolt. Since these AM5 CPUs are power limited, you can get (potentially) much higher clocks while still staying within the same energy budget. It's perfectly safe (AFAIK) since it doesn't involve pushing more energy into your chips (neither voltage nor TDP), and won't take very long.
I'm all for naming and shaming, but why not also name and praise when a company does good by its users?
I'm not exactly in the market for a mobo right now, but I think that's useful information.
I'm not exactly in the market for a mobo right now, but I think that's useful information.
TBH, I'm not sure how much good these anecdotes do. I had bad experiences with all major brands; a high quality & expensive ASUS motherboard was the worst, even though top ASUS is considered the safest choice in circles I tend to spend time in — if you have the money (it caught fire in the middle of light code editing for no reason I could establish after five years of excellent work).
Anyway, I now stick with ASRock. This will now attract anecdotes from users who have run into problems with them, and will describe at length how crappy their products are. Works with literally any brand, so I try not to mention any unless prodded.
Anyway, I now stick with ASRock. This will now attract anecdotes from users who have run into problems with them, and will describe at length how crappy their products are. Works with literally any brand, so I try not to mention any unless prodded.
This one?
https://arstechnica.com/gadgets/2025/07/firmware-update-hind...
The company isn't shutting down, probably just an attempt to get more subscriptions.
The company used to support Strava integration from their app in Free Ride mode. That broke last year, the app still reports that a ride happened, but the ride is reported as 0 miles, 0 minutes long. I have the low end bike and rower from them. The machines are great physically.
https://arstechnica.com/gadgets/2025/07/firmware-update-hind...
The company isn't shutting down, probably just an attempt to get more subscriptions.
The company used to support Strava integration from their app in Free Ride mode. That broke last year, the app still reports that a ride happened, but the ride is reported as 0 miles, 0 minutes long. I have the low end bike and rower from them. The machines are great physically.
How does that work for the PC I just got a month ago that has this motherboard and was advertised as having PCIe 5.0 ? https://www.memoryexpress.com/Products/MX00129652 I just lose capability that was clearly marked as a specific feature in the item I purchased?
Ah, I see that the PCIe 5.0 mentioned in the product specs is for the M.2 storage, not for the PCI slot for a graphics card. The post title should probably say "removes unofficial GPU slot PCIe 5.0 support" or something.
But yeah, considering I have a 5060ti 16gb on this motherboard, I guess I'll be staying with the current BIOS version indefinitely... Just checked with CPU-Z, it says Bus: PCI Express 5.0, current link width: x8 , current link speed: 16GT/s
Ah, I see that the PCIe 5.0 mentioned in the product specs is for the M.2 storage, not for the PCI slot for a graphics card. The post title should probably say "removes unofficial GPU slot PCIe 5.0 support" or something.
But yeah, considering I have a 5060ti 16gb on this motherboard, I guess I'll be staying with the current BIOS version indefinitely... Just checked with CPU-Z, it says Bus: PCI Express 5.0, current link width: x8 , current link speed: 16GT/s
It sounds like this only applies to boards that were never listed as PCIe 5 capable in the first place.
If you read the detailed specs on the page closely it never advertised 5.0 speeds _on the PCIe slots_ only on the M.2 interfaces. The top line product name is deceptive, either intentionally or accidentally.
Full expansion card specs only mention PCIe 4.0: https://www.memoryexpress.com/Products/MX00129652#:~:text=en...
M.2 Storage specs mention the 5.0 speeds and only for one of the slots: https://www.memoryexpress.com/Products/MX00129652#:~:text=pr...
Full expansion card specs only mention PCIe 4.0: https://www.memoryexpress.com/Products/MX00129652#:~:text=en...
M.2 Storage specs mention the 5.0 speeds and only for one of the slots: https://www.memoryexpress.com/Products/MX00129652#:~:text=pr...
It's only for single SSD M.2 storage slot directly connected to the CPU.
Were the lanes engineered for 5.0? Maybe this is a signal integrity issue.
The way product segmentation works in this market it’s practically certain they’re physically identical to the board that supports 5 and the only difference is firmware locks and marketing.
Having some PCB experience, I disagree. Targeting higher specs like PCIe 5.0 usually requires more expensive PCB materials and tighter qualifications at the test stage.
If the boards were only being tested to PCIe 4.0 (their official spec) then you can't guarantee they perform the same as SKUs targeting PCIe 5.0.
You can't notice these differences with your eyes.
If the boards were only being tested to PCIe 4.0 (their official spec) then you can't guarantee they perform the same as SKUs targeting PCIe 5.0.
You can't notice these differences with your eyes.
This motherboard already supports PCIE 5.0 for NVME, so it's likely the board stack up is plenty sufficient for PCIE 5.0.
The differential impedance is also identical for PCIE 4.0/5.0 so again, likely not a physical difference issue.
Which basically suggests that the routed the 4.0 lanes much sloppier than 5.0.
The differential impedance is also identical for PCIE 4.0/5.0 so again, likely not a physical difference issue.
Which basically suggests that the routed the 4.0 lanes much sloppier than 5.0.
It’s not that simple. You could get PCIe 5.0 working on the cheapest FR-4 board if you kept trace lengths short enough.
The trace lengths matter a lot, among other things. You can’t draw conclusions like this.
The trace lengths matter a lot, among other things. You can’t draw conclusions like this.
The general advice is 5 mils Intra-Pair and no specific requirement for Inter-Pair. These follow the same engineering guidelines as PCIe 4.0, with identical impedance and coupling. As far as I can tell, there are no significant differences, aside from PCIe 5.0 having smaller theoretical margins. Even then, the 5 mils spec is likely overkill for both.
From what I’ve read, a well-designed PCB that supports PCIe 4.0 should also meet PCIe 5.0 requirements electrically and for signaling. I suspect the issue may be related more to power delivery or EMC than to trace layout or stack-up. Alternatively, it could just be an AMD policy decision to limit PCIe 5.0 support on this chipset variant, rather than a design flaw.
From what I’ve read, a well-designed PCB that supports PCIe 4.0 should also meet PCIe 5.0 requirements electrically and for signaling. I suspect the issue may be related more to power delivery or EMC than to trace layout or stack-up. Alternatively, it could just be an AMD policy decision to limit PCIe 5.0 support on this chipset variant, rather than a design flaw.
> I suspect the issue may be related more to power delivery or EMC than to trace layout or stack-up.
Power delivery shouldn't be an issue, the slot power limits haven't changed afaik, max is 75W for an x16 card since the beginning.
Power delivery shouldn't be an issue, the slot power limits haven't changed afaik, max is 75W for an x16 card since the beginning.
I'm not sure a PCB of such size can be made (without making the PCB actually more expensive) partially with worse and better materials, since the board does support four PCIe 5.0 lanes. The question is how much cheaper would it be to route the 16 lanes only for 4.0 speed rating.
> since the board does support four PCIe 5.0 lanes
The better materials are necessary for longer traces. You can’t make simple comparisons like this
The better materials are necessary for longer traces. You can’t make simple comparisons like this
PCIe 5 is 32 GT/s (~4GB/s) per lane, and of course double of that of PCIe 4.
It's a crazy rate that can't be trivial to achieve.
It's a crazy rate that can't be trivial to achieve.
It seems pretty worrying, since PCIe 6 for the first time couldn't double the operating frequency and it's not likely to be possible in the feature for traditional copper lanes. I haven't heard how DDR6 will solve this issue. PAM or not PAM?
From guidelines:
"... please use the original title, unless it is misleading or linkbait; don't editorialize."
"... please use the original title, unless it is misleading or linkbait; don't editorialize."
This title appears to be misleading, or at the very least confusing.
You have to pay close attention to the difference between the M.2 slot (which can be PCIe 5.0 on B650) and the expansion slots (which were never advertised as PCI 5.0)
Leaving PCIe 5.0 enabled on the GPU slot could have caused problems if the designs and boards were qualified with the expectation they'd only run PCIe 4.0 speeds. If a board physically can't handle PCIe 5.0 but it gets enabled, it could lead to random crashes and instability, which turns into a higher return rate and angry customers.