News Gigabyte Pulls PCIe 4.0 Support From Pre-X570 Motherboards

[Admin]

Gigabyte releases new firmware with the AGESA 1.0.0.3 ABB microcode that removes PCIe 4.0 functionality from AMD 300-and 400-series motherboards.

Gigabyte Pulls PCIe 4.0 Support From Pre-X570 Motherboards : Read more

 

[jimmysmitty]

It makes sense to remove it. AMD is not going to support it which means a motherboard vendor would be the only contact for any support.

Its better to avoid potential issues rather than enable something and increase potential issues.

 

[linuxdude]

To be honest I was quite shocked when I heard AMD is activly preventing enabling PCIe4 on 300/400-series chipsets.

This is the kind of behavior I am used to see from Intel (extract as much money as possible from customers), but until now not from AMD. It seems success always corrupts in one way or another.

 

[NightHawkRMX]

I understand why AMD would want mobo manufacturers to do this. It creates confusion on what boards support pcie4. Also removing PCIe4 on older boards give another incentive for people to look at X570.

That being said, i don't support AMD doing this. It is a very intel move, and not something i like

 

[waltc3]

From what I read, GB pulled it because it didn't work properly and created WHEA warnings in the process of not functioning. Seems ample reason to pull it, imo.

 

[jimmysmitty]

To be honest I was quite shocked when I heard AMD is activly preventing enabling PCIe4 on 300/400-series chipsets.

This is the kind of behavior I am used to see from Intel (extract as much money as possible from customers), but until now not from AMD. It seems success always corrupts in one way or another.

Wait you mean preventing something that might not work properly and cause you issues is an Intel move?

I understand why AMD would want mobo manufacturers to do this. It creates confusion on what boards support pcie4. Also removing PCIe4 on older boards give another incentive for people to look at X570.

That being said, i don't support AMD doing this. It is a very intel move, and not something i like

AMD is acting like a company. Of course they would prefer you buy a X570 board for your shiny new Ryzen chip. They have shareholders to please after all.

And its not much an Intel move. Intel just tends to build a chipset and CPU around each other. I think thats the best way to go. Besides how often do you need to actually upgrade your CPU these days? If every other year its too much for little to no gains.

From what I read, GB pulled it because it didn't work properly and created WHEA warnings in the process of not functioning. Seems ample reason to pull it, imo.

So instead of allowing something that could cause people potential issues and end up in a lawsuit, since they would have knowingly allowed these issues to persist, they are removing support so people don't have issues. Seems like a proper thing to do to me.

 

[TJ Hooker]

So instead of allowing something that could cause people potential issues and end up in a lawsuit, since they would have knowingly allowed these issues to persist, they are removing support so people don't have issues. Seems like a proper thing to do to me.

How do you figure it could have got them sued? Nowhere (from either AMD or Gigabyte) does it say the 300/400 series boards will support PCIe 4.0. The fact that an option may exist in the BIOS to enable it hardly constitutes a guarantee that it will work.

 

[jimmysmitty]

How do you figure it could have got them sued? Nowhere (from either AMD or Gigabyte) does it say the 300/400 series boards will support PCIe 4.0. The fact that an option may exist in the BIOS to enable it hardly constitutes a guarantee that it will work.

People sue over the dumbest things. I could see it being possible based on them knowing there are potential issues and leaving it enabled. Having knowledge of it causing issues and leaving it on could cause it.

That said I am glad they decided to disable it as there is a good reason for AMD to do so. Backwards compatibility is one thing but pushing a new standard on something that wasn't built for it originally is another and I have almost never seen that in a board ever.

 

[hannibal]

Yep. Good choise! Pci4 is not must. Stable bios on the other hand is. I am sure that there may be some boards that could be stable in long run for Pci4, but most Are not so it is better to get motherboard manufacturers take care of every trouble that can happen if They on their own add that feature back! It can be done ofcourse. But as it has been said in the above comment, it is better have working bios than something that can cause and will cause unexpected problems!

 

[RodroX]

I updated my Gigabyte B450 Gaming X Bios to version F42b, (Agesa 1.0.0.3 ABB) and the results are:

1. The PCI-E Gen 4 option still there to choose it.

2. The high Vcore, high Temps, and high CPU Fan noise, those became a bit better after updating Bios and installing the new Chipset Drivers from AMD page (Of course I still had to manually disable PBO from Bios, which is kinda useless with stock cooler - at least on my case with the air flow I have - and I have to set "Power Supply Idle Control" option to "Typical Current Idle"). This is choosing the "Ryzen Power Balanced Plan"

3. Fun fact is like kinda nothing changed for me, cause I got the same results than before (without updating BIOS nor Drivers), and just setting those options in BIOS plus choosing "Windows Balanced Power Plan"

I think AMD, Gigabyte (in my case) and Microsoft need to put some more work on this to fix this issues.

 

[cryoburner]

Has anyone actually gone through and tested PCIe 4.0 performance and stability on these boards compared to X570? And is there even any consumer-grade hardware currently available that pushes PCIe 4.0 to its limits, as in, close to 8GB/s for an x4 device, or close to 32GB/s for an x16 device?

It's not clear whether this feature would even work reliably on these boards, and in any case, the benefits of PCIe 4.0 for home systems will likely be quite limited for a number of years.

 

[salgado18]

And is there even any consumer-grade hardware currently available that pushes PCIe 4.0 to its limits, as in, close to 8GB/s for an x4 device, or close to 32GB/s for an x16 device?

I've only seen M.2 SSDs that push around 5-6 GB/s. It may take a while for devices to saturate PCIe 4.0.

 

[jimmysmitty]

I've only seen M.2 SSDs that push around 5-6 GB/s. It may take a while for devices to saturate PCIe 4.0.

It would take a few years but PCIe SSDs might be able to saturate a x4 link. GPUs still don't saturate a PCIe x16 link for 3.0 and will be far away from PCIe 4.0. And rumor is PCIe 5.0 will be coming sooner rather than later.

 

[cat1092]

Have never seen any MB OEM 'include' features of newer models, especially PCIe versions. The same would apply with pre-X570 MB's.

That would be like me asking ASRock & ASUS to upgrade my 970 chipset MB's to support PCIe 3.0, which didn't happen. Although both of these OEM's would eventually include 2nd gen (10Gb/sec) NVMe support to some of their latest offerings while waiting on Zen to arrive.

I don't get it, what's there to sue over, false hope? Just sell the current MB & purchase a X570, Newegg is running promos & rebates for some models as of this date. Even the lone mATX offering by ASRock is on promo, along with rebate.

The facts are, these older MB's would be unstable & unsafe by enabling PCIe 4.0 support. Check out the specs of some & one will see why. Most has a fan on the chipset, others has 12-16 pin ATX power connectors versus the usual dual 4 pin. No firmware update can fix the shortcomings of hardware.

However, the one(s) who stated the possibility of PCIe 4.0 support via firmware upgrades should be demoted to a position to where they cannot make statements to the public. That's what this ruckus is about, someone saying it was a possibility. At worst, these comments were an act of stupidity, caused no economic harm to anyone.

As far as comparing this incident to Intel, well they've pulled quite a number of tricks over us, to include not advising Ivy Bridge owners that the IHS wasn't soldered to the die. This would still be largely unknown to many into the Haswell release. It would be writers from tech forums such as this one who exposed the truth over that issue. Later on, it would be someone else (an intern) who uncovered what Intel had swept under the rug years prior, the major security flaws in their CPU's. Sure, AMD also had some, yet nowhere near on the level as Intel.

And still, Gordon Moore hasn't been so much as been indicted by a grand jury for allowing untold millions of those who purchased Intel CPU's with the defects. He will go on to be held unaccountable for Intel's actions under his watch.

This PCIe 4.0 issue is nowhere near on the level as the crimes that Intel has knowingly committed & us end consumers won't receive a cent for our losses.

Lastly, I'm not taking a side, as I've built both AMD & Intel PC's. Although in recent years, due to low pricing, went with AMD 970 series FX-8350/8370 builds. Am looking forward to doing a AMD build, once end of support for the AM4 socket is over sometime in 2020, as they've announced long ago. By then, there'll be another socket altogether, in fact, the 3rd Zen gen may be the last for AM4, as AMD looks to be moving forward faster than Intel these days. Look at long have Intel been promising a 10nm die, while they may have one by now, it's been 4 years in the making. While AMD has released 7nm across most of Zen 3rd gen & being tight lipped over their next move, other than Ryzen 9 3950X due in September of this year.

Cat

 

[cat1092]

It would take a few years but PCIe SSDs might be able to saturate a x4 link. CPUs still don't saturate a PCIe x16 link for 3.0 and will be far away from PCIe 4.0. And rumor is PCIe 5.0 will be coming sooner rather than later.

Yes, PCIe 5.0 will likely be on the shelves in 2021!

As for PCIe 4.0 speeds, am sure that the Samsung 980 PRO/EVO NVMe line will be much faster than any of the current offerings. Likely won't saturate the full spec, just as with PCIe 3.0, yet there'll be great improvement. This is a great time to buy NVMe if staying with PCIe 3.0, I snagged a 512GB Samsung 970 PRO for half the price paid for the 950 PRO of the same size in 2015.

Cat

 

[TJ Hooker]

The facts are, these older MB's would be unstable & unsafe by enabling PCIe 4.0 support. Check out the specs of some & one will see why. Most has a fan on the chipset, others has 12-16 pin ATX power connectors versus the usual dual 4 pin. No firmware update can fix the shortcomings of hardware.

However, the one(s) who stated the possibility of PCIe 4.0 support via firmware upgrades should be demoted to a position to where they cannot make statements to the public. That's what this ruckus is about, someone saying it was a possibility. At worst, these comments were an act of stupidity, caused no economic harm to anyone.

Fan on the chipset is irrelevant, given that PCIe 4.0 on 300/400 series boards would only ever be possible on PCIe lanes straight from the CPU. And those extra CPU power connectors are largely superfluous and just there for marketing, except for the high end boards that are to be used for extreme overclocking of the high core count CPUs. Certainly not in any way required simply for PCIe 4.0. By and large Ryzen 3K chips don't draw any more power than comparable Ryzen 2K chips. Any hardware limitations would be related to PCIe traces and/or redrivers.

As far as "someone saying it was a possibility", that was AMD themselves a while back.

 

[jimmysmitty]

Fan on the chipset is irrelevant, given that PCIe 4.0 on 300/400 series boards would only ever be possible on PCIe lanes straight from the CPU. And those extra CPU power connectors are largely superfluous and just there for marketing, except for the high end boards that are to be used for extreme overclocking of the high core count CPUs. Certainly not in any way required simply for PCIe 4.0. By and large Ryzen 3K chips don't draw any more power than comparable Ryzen 2K chips. Any hardware limitations would be related to PCIe traces and/or redrivers.

As far as "someone saying it was a possibility", that was AMD themselves a while back.

The biggest issue I can see is the traces wouldn';t be designed for it. Most likely PCIe 4.0 has wider traces to account for more bandwidth. While every spec I can find shows PCIe 4.0 to provide the same power at the slot as PCIe 3.0 it would still use slightly more for more bandwidth.

And correct, it was AMD who was stating it. Original statement was they were leaving it up to AiB partners then decided to nix it due to possible issues. I can imagine a lower end board would have had more troubles and not look good.