Not really happy with AM5

Hmm. Early Zen and Zen1+ AM4 Ryzens were very picky about RAM, but this is significantly improved with Zen2 and was almost entirely eliminated with Zen3 and later and is mostly historical at this point. By the time Zen2 came around you could use almost any RAM you wanted without stability issues, and Zen3 and later were even more stable.

Were you trying to populate four RAM slots maybe? Or maybe trying to use dual rank RAM? Because that does run into signaling issues on both AMD and Intel platforms, and usually requires the RAM to run more slowly in order to be stable. Usually the advertised memory clocks are only fully supported when using one single rank memory module per memory channnel. It has been this way for a very long time on both Intel and AMD simply because as memory performance has increased, the signalling has become increasingly sensitive and difficult to maintain.

This is usually documented straight in the manual for the motherboard, but looking at the manual for the ProArt board you have, it seems Asus neglected to mention this. They just reference the Memory support list on their webpage, and it does not speak to differences in memory speed with different module quantities installed. (They do notably not have any 4x kits in the list)

If you were just using two RAM slots, then you may have simply had defective RAM. Sometimes unstable RAM can be made stable by slightly boosting the voltages to the memory or the IMC. (care needs to be taken here to avoid overheating or damage)

It’s also curious that they advertise 8000MT/s capability. While this is true, in most cases when used the performance will actually be lower than slower memory kits. The key to AMD memory performance is to run UCLK (uncore clock, the speed the IMC operates at) = MCLK (memory clock, the clock the memory modules operate at (a 1:1 ratio). This results in the lowest latencies in the overwhelming majority of cases, and bandwidth is directly related to latency.

The issue here is that UCLK has a max specified speed of 3000 Mhz, making DDR5-6000 the fastest officially supported RAM in a 1:1 configuration.

It is possible to go higher (up to 6200 and even 6400) but this involves overclocking the IMC and running it at higher voltages which is not guaranteed to work, will run hotter and if you set the voltage too high can damage things.

If you use higher RAM clocks (higher MCLK) than uncore clocks (UCLK) you will be in an asyncronous multiplier mode, which results in higher latencies, and will be bandwidth limited by the UCLK and even the FCLK (fabric clock, the communication link between the CPU dies and the IO Die) and this is generally not positive for performance, but there are exceptions with certain very specific workloads which have to clear the memory often, in which case the higher clocks reduce the wait time while the memory is cleared.

Absolutely.

But unfortunately what we used to call HEDT is dead.

The last were probably Intel’s x299 and AMD’s Zen2 Threadrippers (Threadripper 3xxx)

After that generation HEDT disappeared, and instead you got a choice between proper workstation or consumer product. The in-between “pro-sumer” category pretty much disappeared. And I think there are many reasons for this.

1.) Most people who bought HEDT platforms were high end enthusiasts, but as prices for consumer platforms started getting higher and higher, there was a diminishing number of people who were willing to spend more than what a high end consumer platform cost, other than the people who were already buying proper workstations already, and as volumes diminish, costs further increase making it even less viable.

2.) In the past with DDR4 some systems were able to be compatible with both UDIMM’s and RDIMM’s. The flexibbility allowed HEDT platforms to flourish, with high end prosumers installing cheaper (and faster) UDIMM’s and light workstation users able to install RDIMM’s. With DDR5 this is no longer possible which forces a delineation. Most workstation products require RDIMM’s and consumer products require UDIMM’s. This helped kill off the tweener HEDT platforms as the flexible crossover between the two that allowed for greater volumes justifying the platform was now gone.

3.) Increased internal cannibalization of higher margin products by lower margin ones. If a consumer platform gets more than 2 memory lanes, or gets more than 28 PCIe lanes, then some professional users may start buying cheaper consumer platforms instead of buying higher margin professional platforms.

Sadly, I think that era is now thoroughly dead. The market is now thoroughly delineated. There are not consumer platforms with high clocked CPU’s with high performance per thread and low latency RAM which are great at consumer applications and games, but suffer from having rather few PCIe lanes and few memory channels, and there are workstations which now are more like their server counterparts, highly capable with many cores, lots of memory channels and lots of PCIe lanes but designed with lower core clocks for lower performance in lightly threaded tasks, and higher latency buffered RAM.

I don’t think this will change. The costs are too high, and the benefits to both the board makers and chip makers are too small and the risk of cannibalization is too great.

What this means is that even though AMD has been making 16C/32T consumer CPU’s for some time now, for many workloads there just isn’t enough memory bandwidth to keep up when all cores are busy, forcing them to sit idle a lot waiting for RAM, and that harms performance.

The jump from DDR4 to DDR5 did help things though by roughly doubling the available RAM bandwidth, but then again, in that time the CPU cores also got faster and gobbled up a lot of that increase.

I’m with you. I have been lamenting this loss for years now. It is the main reason why I am still using my Threadripper 3960x as my main workstation, simply because modern proper workstation platforms are crazy expensive, and modern consumer platforms just don’t have enough PCIe lanes for my needs.

My Threadripper 3960x like my x79 system before it and my x58 system before that all served as “one system to rule them all” builds with no compromises. Able to both be top tier platforms for work and expansion and for the rare game I play at the same time. Now we have to choose.

When the Threadripper 3960x started to become a little slow for some game titles I wanted to play I made the difficult decision to move to a two system setup, keeping the Threadripper for work and expansion, and building a dedicated AM5 system for when I want to run a game. I hate that I have to do this, but that’s the world we now live in.

I feel your pain. My Asus X670E ProArt has been a giant pile of fecal matter the entire time I’ve owned it:

  • Flaky onboard 10Gbit (currently disabled)
  • Sometimes boots and OS is extremely choppy and unusable
  • Sometimes boots and primary PCIe is stuck at x3.0 speed
  • Sometimes boots, crashes, and restarts itself
  • Sometimes boots and works perfectly forever until shutdown/restart/suspend
  • ASUS BIOS flash utility hasn’t worked for the past 4 or so BIOS updates. Just hangs on reboot and never flashes. Have to use the BIOS flashback button.

I’d rip this turd out and replace it, but every single X870E board has some kind of major deal killer, be it bad slot spacing, bad PCIe lane splitting, bad onboard Ethernet (dual 2.5G or 5G on a $500+ board is unacceptable), or inadequate USB port selection. So glad I went with Arrow Lake for my office PC.

Wow. Asus used to be regarded as one of the highest quality brands in the industry.

My last Asus motherboard I had was my x79 system with my Core i7-3930k, and it was a rock solid product I was very happy with, but I bought that 15 years ago now.

Back in the day the consensus was that Asus builds some of the best products in the industry that rarely fail, but you had better hope that they don’t fail, because their RMA process was one of the worst in the industry.

Do you think you just got a lemon, or has Asus overall gone to shit?

Fun facts…

EVGA had the ultimate overclocking and highest end boards for the X58 (FTW and Classified) circa 2008-2010.

Then Asus paid an obscene amount of money to hire the BIOS team lead (Shamino, aka Andre Yang) and fellow teammates away from EVGA to Asus in Tiawan, where they released the Asus X79 RIVE and then the X99. EVGA’s X79 had issues for LN2 overclocking, and Asus took the crown. It was all about the BIOS Team.

Then they retired wealthy enough to live the rest of their lives. Giving Asus the big finger. So we lost both EVGA and Asus top end boards within a 5 year span and overclocking has never been the same since their magic.

Ive never had as much luck with Asus ever since my X79 RIVE Black Edition.

ASRock (a subsidiary of Asus) on the other hand has surprised meany times in both consumer and enterprise motherboards with their BIOS and BMC implementations. its been the only brand I’ve purchased since the X399 Tachi/Epyc Gen1 days to present B850 and WRX80 systems.

1 Like

My setup is using a ProArt x870e. There was some lane fiddling to get the 10gig LAN working at the same time as 2 GPUs but it’s stable and really doesn’t have any issues for my use case.

That’s the main thing for me I think. For me at least, I was still able to build a crazy 16 core 256gb machine with 3 video cards and two high speed storage devices all on dedicated lanes, unfortunately only 5gb lan and I had to sacrifice some USB to achieve that, but it is what it is, and because that exists, they’ve already lost what would have otherwise been a workstation/hedt customer. Pushing the consumer platform any higher means they lose even more workstation customers and it just doesn’t make sense financially.

I’ve still got an EVGA Z170 classified K in one of my machines, it has a built in PLX switch for 4 way SLI, CPU sends x16 to the switch, which splits it to x8 for each of the 4 GPUs and sends each the same data. Probably not much real world application now, but you don’t see cool stuff like that around anymore.

1 Like

How did you get 3 video cards on dedicated lanes, on a platform that only has 28 PCIe lanes in total, 4 of which are reserved for the chipset?

This thread got me curious so I peeked at the current LGA1851 options (Microcenter has some good bundles). Intel has basically the same lane constrictions. :frowning:

I’d say Intel is ahead here, they have about twice as many useable chipset PCIe lanes as AMD does due to having double the CPU-chipset interconnect bandwidth and not forcing the motherboard manufacturers into allocating PCIe lanes into (imo) less than useful USB configurations.

Maybe I looked in a wrong spot, but doesn’t Intel have 4 less lanes? AM5 spec is 28 and LGA1851 is 24.

They both give you 1 x16 slot, 1 storage slot, and the rest is YMMV based on the board.

Indeed. While they look the same on paper, the big distinction is that while AMD’s terminology counts the chipset link as PCIe lanes, Intel calls those DMI (Direct Media Interface) links. While they are essentially just PCIe lanes with some extra flow control (I think) since they use a different term for them, they are not counted in the PCIe lane totals.

As of right now, I believe current DMI implementations also have a little bit more bandwidth than AMD’s 4x Gen4 chipset links, meaning you can get more chipset lanes before you run into bandwidth constrictions. (though in reality I have not seen things implemented that way in the boards I have shopped for, which may make it a little bit moot.)

I would love to see AMD up the chipset lanes to Gen5 since the CPU already supports it, and provide more options for chipset lanes, but they aren’t exactly incentivized to do so, if it makes consumer platforms more viable for people who otherwise would have bought their workstation platforms unless they think they are losing sales to Intel as a result, which I am not sure is the case.

1 Like

AMD and Intel aren’t aligned in what they report here. AMD includes it’s chipset connection in it’s PCIe lane count while Intel does not. Intel’s chipset connection actually has twice as much bandwidth as AMD’s but it isn’t often mentioned in marketing. Also their chipset PCIe lanes are much better implemented than AMD’s aging Asmedia chipset.

Ah! Good to know! Thanks for the clarification. :slight_smile:

They’re not all direct to CPU. The MSI carbon max supports two gen5 in x8, and then has a separate chipset controlled gen4 x4 that I’m running my 3090 on for display out, 3d acceleration, and vram spillover. Works great. Also you mentioned less than useful USB configurations, the board also allows you to get full bandwidth on the second ssd by switching off some usb4 ports. Interesting idea.

Just want to correct something here.
Shamino is definitely not Andre Yang.
They’re two different people.

From what I understand, the difference is things like non-PCB material choice, BOM components, convenience features (e.g. onboard Power and Reset buttons), etc. It’s true that PCB tends to be universal, but the components on the board matter too.

Higher end ASUS boards have advanced features like SlimSAS PCIe 4.0x4 ports that don’t share lanes with the PCIe 5.0x16 slot. This allows a consumer board to support enterprise hardware like U.2 SSDs, and in fact yes there are people in the ASUS DIY community who do exactly that.

Videocardz have had the best coverage of this and haven’t posted anything about it in a while. The problem was mostly an ASRock one. I saw 1 headline about Asus (I think) but nothing else has cropped up since. Of course, that could also be because the number of people actively building PCs at any given time has fallen off a cliff due to memory pricing.

From my observation, the common thread was waking from sleep, which AMD CPUs are notoriously bad at. My guess is the board was supplying too much power to CPU too soon.

So far so good with 9950X3D2 + ASUS ROG CROSSHAIR X870E HERO build. I have sleep disabled.

J2C’s 285K video bears this out.

Granite Ridge is a viable solution-ish if you’re not chasing framerates.

TIL, thanks! The CPU spec sheets definitely make it seem AMD has more lanes. That said, I did build an AMD PC as I didn’t have any other AMD desktop and wanted to get some hands on experience with that. OTOH I have both Intel and AMD laptops.

For observable differences I’d start with a close look at the DIMM sockets, maybe who’s making the AM5 socket as well, and the extended timings in MSI BIOSes.

Besides dielectric, there’s possibly some higher BOM allocation to buried vias. But checking that takes stripping multiple boards. Surface finish on the striplines could also be smoother and details of FR4 stranding might shift, though that’s more knowing what’s going into the PCB assembly.

AM5 works just fine in general, when you start to go out of spec in one way or another that’s when you run into issues in the majority of cases.

…and of course, as with any platform if you go for the bottom of the barrel motherboards you’re more likely to run into issues but that’s not news specific to AM5.

I have two AMD rigs, one with a MSI X870E 9950x3d 64GB DDR5 6000 CL30 (2x32GB) and has been rock stable not one problem. I use GSkill and Kingston Fury memory.

MSI X670E 9800x3d 64GB DDR5 6000 CL30(2x32GB) rock stable.

memory matters which brand and make sure you setup them correctly in the BIOS.