Updated: Jul 6
For me, in my tiny minds eye at any rate, I concluded that Samsung are sort of to blame for the bad electronic device Karma we are all experiencing right now, as electronics everywhere seems to be suffering from a horde of global virus like gadget crippling symptoms and one would be forgiven for idle speculation that they are maybe suffering various maladies and shortcomings in symbiotic sympathy of the human condition of the moment...?
The way I saw it, the Samsung S20 series debacle started it all back in 2020 (In My Holistic Opinion) with the horde of craptastic components finding their way into USB-C hubs, defective cables, shortsighted computer OEMs ignoring the unignorable and the very obvious consequences that their deliberate ignorance of this self inflicted 💩💩💩 pie would bring seemed pretty obvious to moi.
I had strawberry jam topped scones with Dirk Gently from the holistic detective agency the other day and while I was sipping a fine Darjeeling tea and swapping notes to this end with the chap he just disappeared into thin air on the third revolution of his teaspoon as he was stirring his Earl Grey tea just as we came to a mutual consensus on the matter.
I blame the cherry blossom honey he puts into his Earl Grey but he does this sort of thing all the time......Hard to tell what's what around the guy....
Sadly, it has turned out exactly as we both suspected.......
Maybe I play too much Chess but it struck me there could only be one possible outcome to this whole USB-C hub saga thang??
You do not even have to be particularly bright to have figured that one out!!
This all combined to literally set 🔥🔥 to the trend that got this gadget malady theme really pumping iron and revealed that Intel were not alone in a series of serious chippery based glitches and that AMD and Apple (amongst others) are suffering from these challenges as well.
I did actually pick up on this latest AMD issue back when I was using my X370's which saw me plug all my USB devices into the USB 3.0 ports as a simple remedy and I thought nothing more of it other than interaction I was having with AMD about BIOS, Ryzen Master system device manager control software glitches and XMP memory clocking challenges.
I meant to bring up this minor USB Port irritation with them but never did.
This is not a new issue either by the way as I remember ignoring various USB ports and they were the USB 2.0 ports on the X370 series motherboards as well.
Seems I was hiding the issue denial style and eliminating the obvious by doing this.
It is funny how many times we all do that when caught up in the more complex issues of the moment.
I remember a few years ago a pal of mine brought me a sick Kawasaki KLR650 to diagnose.
I swiftly ascertained the electronic ignition system was faulty and replaced it with the cheaper but exact same component from the KLR250 series.
That part I got absolutely correct.
Kawasaki themselves later verified the EI unit was kaput.
Then the freaking thing was miss firing and I wrongly concluded that the carburetor was at fault for that malady but after ripping it apart and destroying the thing was embarrassed by the motorcycle mechanic who told me immediately it was the spark plug after I shared with him the carburetor had no issues before I broke it.
It was indeed, the darn spark plug......
I subsequently fixed many sick motorcycles with a mere new Spark plug set....
The new carburetor was $450, a new NGK spark plug was $2.75.Enough said..... 😩😭
I often find simple problems with computers and their cables as the sole issue with an entire data Center going awry in a similar vein.
I actually dismissed an AGESA issue on the X570 as impossible as it was just far too simple a diagnosis to be possible!?
I was reading an article on March 12th that exposed this AGESA issue that has beset our pals at AMD and started linking and joining the dots together on my own systems.
Almost all of my AMD compute devices (7 now) display this USB port characteristic but it does not affect the compute part of what they do because I only hang USB based video conference cameras and microphones off them now.
Stuff I used to plug in (but no longer do) to USB was a keyboard and mouse which are now all Bluetooth devices.
Less on USB is better is my new rig mantra!
I have also been diving into the AMD AGESA "fix" only to discover they are doing these the Intel way (workaround, not a real fix).
I do not think they have a choice here either actually as the glitch seems to be in the silicon.
So I am doing my own workaround for this problem that will make the USB 3.2 slots useful for just bootable USB purposes and Video Conferencing gadget purposes only.
This sad saga by the way is sorta related to the general mess we have with USB-C hubs & Cables and why they are now a pretty huge problem.
New BIOS revisions that address this latest AMD AGESA USB 2.0 issue are already rolling out though.
I have already downloaded and installed several iterations of these already as all bar one of my devices use Gigabyte TRX40 or Gigabyte AM4 X570 motherboards.
The issue becomes noticeable when you install any newer PCIe 4.0 nVidia GPU into your system.
I do not use nVidia PCIe 4.0 GPU on my AMD fare these days as a result.
The Trump tariff thang has had a bad impact on computer builders like myself with Video Cards doubling in price by the way.
I noticed this while looking for pricing on basic ones this past weekend I thought I had woken up in an alternate reality when I looked at their pricing!!
Online technology comics report a few hundred dollars increase but you try order one - all you get is out of stock messages everywhere!!
SSD is going to be impacted with this issue fairly soon as well by the way, so do not dither, get your PCIe 4.0 Gen 2 SSD now!!
I have spoken!!!
This actually makes no sense as last time I looked at a map, Taiwan was not a part of Communist China, and that SSD chippery stuff is mostly made in Taiwan foundries anyways but what do I know?
We will no doubt hear stories that the things are assembled in Shenzen or something like that....
My newer GPUs are all Radeon 5600 XT series PCIe 4.0 fare and I am not a real heavy GPU user apart from on my one Gaming rig but bizarrely that Rig is not having any issues but it is not using the Gigabyte X570 MOBO either.
I sold the 6800's for a fortune and got 5600's at the old price.
All my CPU's get whacked by my AI software though. All cores all the time.
On my Gaming rig which I play Crysis on, I do not see this behavior.
I went with AS Rock X570 Aqua on that baby.
I am also about to switch to the new WD_Black SN850 series as this makes a huge difference to overall rig performance and the price is now right so I may see bigger problems as I load up on more PCIe 4.0 peripherals.
The WD_Black SN850 500GB comes with a 5 year warranty and the price is pretty sweet.
I went this way because I already have too many NVMe Gen 3 SSD that are just fine and do not want to not use them.
Be careful when you buy the Sabrent SSD by the way, the ones that look like a bargain are but they are using the Phison E16 SSD controller. These are Gen 1 PCIe 4.0 SSD chips.
You will be disappointed if you expect the Gen 1 SSD to benchmark like the Gen 2 stuff but they are still cheap and still very fast.
They are still PDQ compared to the very best PCIe 3.0 NVMe goodies.
Now, if you have an X570 but are still using PCIe 3.0 devices in it you may just notice some of the USB 2.0 Ports on your motherboard suck, so use the white, blue or red ones instead.
In fact blank all them USB 2.0 ports off like I did and do not use them at all, even when the BIOS fix comes out and the new chipset drivers AMD are working on right now become available.
This is what I did to eliminate the problem for all of my rigs even on my X370 fare that last years lightning storm toasted but I did not dive into it much then as I was aware from my Jaguar EV400 HSE adventure with USB-C cables in a car with USB-A 1.0 Ports that things here are not kosher anyways.
Compared to the more serious DIMM and meltdown issues I have been diving into on Xeon fare, a mere AMD AGESA USB port issue seemed like small potatoes for a bug.
It is and it isn't.
Happily there are reasonable workarounds.
More of that later.
Just so you know, the SSD world has changed from the usual players and now we find Sabrent and WD are ruling the SSD roost with Samsung stuff being a pretty distant 3rd in the rankings.
Phison are working on their E21 through 24 series stuff NOW and some preliminary PCIe 5.0 goodies in the lab.
I hear the E24 stuff is like dark arts magic in comparison to the latest and greatest E18 fare.......
I am sure Samsung will reclaim their top spot in short order but right now Phison based SSD chippery rules the roost.
The only advantage that Samsung has right now is that they still do the 500GB M2 NVMe PCIe 4.0 based stuff for a reasonable outlay while most of the the other new players are mostly doing 1, 2 & 4 TB M2 PCIe 4.0 NVMe 2240 like SSD only.
This sizing thang is actually quite an interesting topic in it's own right.
The Samsung PCIe 4.0 controller is slower than the Phison stuff by the way...
Some of the newer X570 motherboards are sporting up to SIX M2 slots!!
However, even with PCIe 4.0 you really only have one slot capable of max performance at ultra low latency and that is slot A that talks directly to the CPU.
Slot B is on a different bus.
In the AMD Universe that's only slightly slower by the way, they do not have the crippling bus issues Intel still have because Intel never solved them properly anyways as that would have required a brand new chip architecture they felt was too costly to get into.
I wonder if they lament the fact that they could not afford not to design their user desktop chips properly given what AMD is doing to them right now....
I hear some Intel board members suggestion to close the gap was to buy AMD lock stock and barrel...........
An admission of defeat if ever I heard an admission of defeat!!
Back to the SSD strategy to populate a X570 motherboard with, which right now are to the advantage of the consumer.
You will find that you have many options to suit your budget and cutting edge gaming aspirations these dayz.
These market changes with OEM SSD vendors and the huge availability of old and new SSD out there is such that it would make sense to have one large SSD in slot A and use the older PCIe 3.0 or even the slower Gen 1 PCIe 4.0 SSD.
If you have hordes of older Samsung SSD like I do put them to good use, you do not need to get rid of them!!
I am eyeing out that motherboard with the six NVMe SSD slots as I can populate them all with different generations of SSD I have lying around doing nothing.
I am thinking of building a Super CAD workstation rig with dual Radeon 6900 XT GPUs and dual Samsung 8K monitors and a Ryzen 5950X CPU in it with assorted SSD.
Seems like 128GB of 3600 RAM will be required for this thing as well with Windows 10 Pro for Workstations.
The price of the 8K monitors is putting me off this idea somewhat but you never know....
In one of my X570 rigs I am testing multiple generations of SSD I have lying around and that one is using the WD Black PCIe 4.0 SN 850 as the boot and OS host drive, my old Samsung 970 NVMe PCIe 3.0 SSD as drive D: and my older Samsung EVO 850 SATA SSD as drive E:
Seems to be working pretty well and I am Syncing my Synology DS918+ NAS device drive data to the EVO 850 as well so I have backups of my hot active working data on that rig.
I will do the same on all the other X570 rigs bar the Crysis gaming machine so I will have one master and four backups.
RAID setup combinations with a pair of exact drives have some interesting and quirky configs but are not really true RAID due to the fact (as stated) that slot A is directly on the CPU bus while the rest are using the peripheral bus.
Most of us will never notice the difference though.
Another thing to consider here is do you really need to have the latest and greatest expensive SSD PCIe 4.0 Gen 2 stuff when actually the Gen 1 fare is still unbelievably fast (and real affordable) for your general compute needs?
Really, if you are going to be on the bleeding edge here you need to go hogs wild and do it to all the other components in your rig as well.
This starts with 32GB of 3600 DDR RAM.
This has issues on X570 and TRX40 though so I do not advise it, 3200 DDR RAM is very stable and affordable. Bleeding edge does involve some bleeding........bugs, $$, angst et al.
Pick your poison(s)....
Then you go Sabrent Rocket 4 Plus 1TB, 2TB or 4TB or the WD_Black SN850 stuff which does start at 500GB by the way and then slot in a Radeon 5600 XT GPU or two and you are done.
If you want Ulta Bleeding edge GPU the AMD Radeon RX 6900 XT is for you but good luck getting your grubby paws on one!!
The new Gen GPU cards are much cheaper than the previous gen because the chippery is being mass produced now.
Do not buy a GPU card based on price alone!!
Whatever you do, make sure your new PCIe 4.0 GPU has 8 GB DDR6 RAM on it!!
This will require you to set PCIe 4.0 support to on in your BIOS which on mine have always been on anyway.
This is where the AGESA based USB fun WILL start.
Plug your most important device into the USB 3.1 or 3.2 Ports (Keyboard or mouse) first and ignore the bank of PCIe 2.0 ports at all costs for now.
April/May 2021 saw better BIOS revisions to cure this malady or make it better.
For those of you thinking this is to the advantage of getting Intel based chippery, I will do another blog that will reveal other Intel issues are far worse and deeper than this mere USB one.
This AGESA USB 2.0 issue is in fact nothing in comparison.
AMD Ryzen is still the best compute octane available by a long stretch and new Intel Rocket Lake fare per whispers I am getting from my Intel gurus is still not fully PCIe 4.0 compliant.
It will run PCIe 4.0 SSD but not PCIe 4.0 GPU??
Oh the things I wish I did not know!!
Also, PCIe 5.0 stuff is going to be with us relatively quickly. There is not going to be the many years lag we saw between PCIe 3.0 and PCIe 4.0.
PCIe 5.0 is in fact, ready to roll right now.
The only question to ask is what is Intel doing about full PCIe 4.0 capability and what are they doing about PCIe 5.0 futures?
The answer is not as much as what AMD are doing. Not by a long chalk.
The reality is that AMD are in another dimension here, not just a country mile or three ahead of Intel. They are in a totally different dimension altogether.
In fact Intel is in danger of slipping to third spot in the CPU stakes as ARM based chippery in computers are now making an appearance and their performance is blistering.
It also takes the GPU into another dimension while it is at it and this new chippery is also all 5nm!
The new ARM cortex X1 Core married to the ARM Mali-G78 GPU has eye popping 40% more GPU performance increase range to slobber and drool over.
I heard Apple are waiting for PCIe 6 before they ramp up from PCIe 3.0, that is soo boring it ain't funny.
We are now in the zone where serious AI computing has no barrier.
The signal processors these things come with gives you 200 megapixels to play with on the camera sensor front.
The ARM players here are Apple, Qualcom (Snapdragon) and Exynos (2100).
Samsung announced the S21 will use both Snapdragon 888 Fare and the Exynos 2100 chippery.
By the way nVidia bought ARM recently.
A very smart move actually.
I think the new boss at Intel, old Pat Gelsinger who was at VMware and before that EMC has some serious catching up to do at this juncture.
He may just be on an exercise to enhance futility itself there....
Intel did not put enough R&D into the right places to make Optane a relevant technology with their sillycon.
This is not the kind of gap you close in 18-36 months either.
At least Pat understands all too well what R&D did for Evil Machine Corp all these years.
I am confident with him at the helm that the Intel foundry will be cooking up some new chippery, it is more of a prayer than a reality though....
Nod off in this biz like Intel did and you find yourself swamped by the new wannabees who are very happy to slit your corporate throat while they whizz merrily bye.
It's brutal competitive out there these dayz.
If Pat was smart he would form a co-operative of sorts with AMD, kind of like the way VCE was created between VMware, Cisco and Evil Machine Corp.
Now that would be interesting.
In the meantime, Intel needs to start work on building a new decent CPU architecture free of all the x86 and x64 baggage that crud comes with.
The Itanic was supposed to blaze the path here but $60 billion plus invested in that made it the showcase example on how to waste investment dollar.
I blame HP and Intel equally for that debacle.
The Good ship Itanic
They (Intel) also need to focus on 5nm or 2nm chippery know how on the manufacturing front, 2 nano is coming!1
IBM showcased their new 2 nano chippery the other day, so it is not science fiction fantasy, that shit is here!!
They need to innovate and make that Optane investment pay in the CPU domain.
In a lot of ways Intel has become like the US Post office - old, stale and fuddy duddy with inflexibility to move with the times a key character trait on display.
This might require plenty of new breaths of fresh air in the corridors and offices at Chipzilla to counter?!
I often wonder what's stopping Intel doing their own ARM chippery path in any event??
Seems moving from making Sillycon to Silicon might be in order?
If you can't beat them join them, what, what?!
Apple just did it!!
Things to wonder about in 2021 and beyond...?
So now on to workarounds to this AGESA USB 2.0 malady.
I already discussed one of them, plugging your USB critical items into the Faster USB 3.1 or 3.2 ports.
The other thing I did was install a Bluetooth PCIe card in all my AM4 rigs that did not sport one on the mobo.
This means USB mice and keyboards can be free of the potential stuttering on the device.
The other thing you can do is replace USB based storage devices with ones that use Ethernet.
I have a Synology DS918+ armed with 2 x 256GB Cache SSD and 4 x Seagate 3 TB HDD to give me 8.5TB of usable disk storage on 1 GB Ethernet.
It's not as fast as some of the USB devices but it don't hang my systems either.
All my systems sync with this NAS and have their own copy of the Data on my SSD archive disk on each system.
I had to put the Logitech Unifying Receiver in the fast USB 3.2 Slot on one of the Ryzen X570 systems and go back to the Microsoft wired keyboard as the K380 was not working anymore after the F33c BIOS update and new AMD Chipset drivers were loaded up.
This Logitech issue was fixed in F33h by the way.
These are just interim steps to the better cure which is coming in April 2021.
As I saw with my Logitech keyboard not working on one X570 Elite, I do not think this actually cured anything.
However, what I did notice even with the XMP off in the BIOS after boot (annoying it does not remember the last BIOS settings) was that the systems all seemed much snappier in response.
So much so that it was only when I bench-marked them all again that I realized I had to go turn XMP back on in all of them as the memory tests sucked ass.
I myself could not tell.
I am also finding I have to do the Afterburner and Kombustor thang as well for the GPU overclocking on each rig which is annoying.
This is not really over-clocking though.
GPU's standard out the box are de-tuned to around 17% of their potential capability for reasons I never was able to ascertain and all this afterburner rigmarole does is push it to it's more natural near 100% setting.
This whole GPU tuning & settings business I find as bizarre as the whole USB-C hub caper thang actually.
I also connect all my printers wirelessly and all my computers use 1GB Ethernet so I am not having much USB traffic in any event.
Strange things happen when I push various USB keys into the free 2.0 ports though which is why I blanked them off.
Some folks visiting were plugging in USB sticks of their own and upsetting the rig somewhat.....
Just one parting FYI on this subject is all Gigabyte X570 and TRX40 motherboards have one special port for BIOS and USB OS load.
I found Samsung metal USB sticks worked best and got a pile of them.
Oh and I also just noted that these USB issues are not happening when these systems run Mint Linux 20.10 Mate on them?
I need to mull that one over for a bit.
I may have jumped to the wrong conclusion there but it sure looks that way to me.....
No actually, the Linux systems logs show the problem is just as bad, it was just hiding it better.
So I just mapped out all my MOBO's USB 2.0 Ports and blanked them off with metal tape I use to seal wing joints on aircraft to eliminate most parasitic drag points.
The Systems are now all also using Bluetooth for Mice and keyboards, just Camera units and microphones for video conferencing is on the USB 3.2 ports.
I thought my X570 rigs were USB 3.1 but that was the version one ELITE motherboard.
My horde has USB 3.2 as standard and all 2.0 ports are blanked off.
My systems are all now much crisper in response since I blanked off the USB 2.0 ports.
I also made sure the front ports were USB 3.2.
My WD_Black SSD order turned up and they are all running real swell apart from the fact they are not hitting 7000. I get 6800 out of them on reads and 5244 on writes.
I suspect this USB problem will not be fully cured until the April 2021 BIOS fixes start their osmosis bleed into the general computing environment but Gigabyte are all over this one and I even heard of some spirited shouting matches going on between AMD and Gigabyte over this very embarrassing issue.
This by the way is also why some of these fancy nVidia 3000 series GPU seemed to stutter and splutter...
Trust me, for AMD it does not get more embarrassing than this sh1t....
I wonder how I would go about teaching them the radical concepts and principles behind testing and Q&A??
Oh the Angst I do not miss!!
April 3rd 2021 Update on AMD X570 AGESA and BIOS updates
OK, so on April 1 or thereabouts, Gigglebyte posted a leap from F33e on the X570 Elite to F33g and on the Chipset driver front there were updates as well as Ryzen Master updates to .118.
The TRX40 Aorus Xtreme BIOS is F4p but I hear there are several updates in the old mixer coming our way for those bad boys as well.
The X570 stuff seems full of frantic activity on both the AMD side and the Gigglebyte side of the equation here.
I myself am not giggling a whole bunch......
Especially after I spent most of Sunday rebuilding one of my Ryzen boxes when Aomei Partition manager deleted the 100MB GPT Boot partition after I got a warning it was in a critical state.
The interesting thing is I had another do the same thing and Aomei handled that one AOK.
I lost all data on the one system's hard drive and then Windows would not install on it to add insult to injury.
I had to install Mate Linux on it first.
Five hours of my life I ain't getting back............was fun though!
I have actually done this Linux install trick a few times before to overcome this Windows inability to get it right.
In fact I must conclude that some people in Redmond are not very clued up about MBR and GPT sh1t from what I saw here with this issue.
I have had this issue come up on a few of the exact same machines every time and my question is why does it not happen on the other two?
I ran install twice after it ran into bizarre issues on this one and I had to use the Boot USB slot on the X570 Elite as well.
Dunno why it worked the third time I did the exact same thing. 3rd time lucky I guess?
The hundreds of messages before that about it being unable to install on a GPT disk even after I ran all the diskpart commands and clear disk and volume deletes etc. were a tad bizarre, it were!
It's an X570, EUFI is its natural state, CSM was on, it would not boot from USB with it disabled??
Come on Microsoft, if Linux can get it right what's the problem?
Now the Huawei laptop that was the problem child is the star of my lineup?
The Universe is running amok again...Great leapin lizards Larry!!