Investigating AMD’s Latest AM5 AGESA Updates on ASRock’s X670E Taichi

It is secure to say that the final couple of weeks have been a bit chaotic for AMD and its motherboard companions. Sadly, it has been much more chaotic for some customers with AMD’s Ryzen 7000X3D processors. There have been a number of experiences of Ryzen 7000 processors burning up in motherboards, and in some instances, burning out the chip socket itself and taking the motherboard with it.

Over the previous few weeks, we have coated the problem because it’s unfolded, with AMD releasing two official statements and motherboard distributors scrambling to make sure their customers have been updating firmware in what seems like a grab-it-quick fireplace sale, pun very a lot meant. Not every thing has been going in accordance with plan, with AMD having launched two new AGESA firmware updates by means of its motherboard companions to attempt to tackle the problems inside per week.

The primary firmware replace made out there to distributors, AGESA 1.0.0.6, addressed experiences of SoC voltages being too excessive. This AGESA model put restrictions in place to restrict that voltage to 1.30 V, and was rapidly distributed to all of AMD’s companions. Extra lately, motherboard distributors have pushed out even newer BIOSes which embrace AMD’s AGESA 1.0.0.7 (BETA) replace. With much more safety-related modifications made beneath the hood, that is the firmware replace AMD and their motherboard companions are pushing customers to put in to alleviate the problems – and stop new ones from occurring.

On this article, we’ll be having a look on the results of all three units of firmware (AGESA 1.0.0.5c – 7) operating on our ASRock X670E Taichi motherboard. The aim is to uncover what, if any, modifications there are to variables utilizing the AMD Ryzen 9 7950X3D, together with SoC voltages and present drawn beneath intensive reminiscence based mostly workloads.

Right here is our current protection of the Ryzen 7000X3D/7000 ‘burnout’ points, together with two statements from AMD, an official response from ASUS and MSI:

AMD Ryzen 7000 AGESA Firmware: From 1.0.0.5c to 1.0.0.7 Inside 32 Days

The primary firmware replace made out there to distributors, AGESA 1.0.0.6, addressed experiences of SoC voltages being too excessive, with new restrictions put in place to restrict issues to 1.30 V. Within the case of the board we have been utilizing to attempt to dig deeper into points, the ASRock X670E Taichi, this was made out there to the general public on 4/27/23 by means of its 1.21 firmware replace. Extra lately, on 5/4/23, ASRock made its newest 1.24.AS02 firmware, which incorporates AMD’s AGESA 1.0.0.7 (BETA) replace.

The AGESA 1.0.0.7 (BETA) replace is the firmware that AMD has been planning to roll out most lately to alleviate the problems of burnout, not only for Ryzen 7000X3D chips with 3D V-Cache, but in addition throughout the broader Ryzen 7000 and AM5 ecosystem. In between the preliminary AGESA 1.0.0.5c firmware that introduced AMD’s Ryzen 7000X3D help to AM5 motherboards, within the house of a mere 32 days, AMD has launched a complete of three main AGESA variations, which ASRock has dutifully revealed for the X670E Taichi. We’ll be utilizing this as our baseline for our evaluation and look into what is going on on.

On high of this, AMD can also be planning to launch an much more robustly up to date AGESA firmware, which may very well be within the coming weeks. Referred to internally as AGESA 1.0.0.9, we did attain out to AMD for touch upon this, however our rep could not touch upon “unannounced or inner solely software program stacks.” It must also be famous that the present firmware on the time of writing out there to customers is a BETA model, implying {that a} newer AGESA is undoubtedly on its approach. Nonetheless, the timescale of the discharge is anybody’s guess at the moment.

So wanting on the variations in AMD’s AGESA updates during the last month, there hasn’t been any official indication of modifications aside from the naked minimal, a minimum of not from ASRock’s descriptions. The next is what ASRock is saying in regards to the descriptions of the AGESA updates:

• AGESA 1.0.0.5c: Preliminary help for Ryzen 7000X3D processors with 3D V-Cache.
• AGESA 1.0.0.6 (BETA): Improved reminiscence compatibility, Optimizations for Ryzen 7000X3D, really useful replace for Ryzen 7000X3D processors.
• AGESA 1.0.0.7 (BETA): Assist for 48/24GB DDR5 reminiscence modules.

The outline of the modifications, a minimum of from the purpose of ascertaining what every AGESA is providing, is borderline pitiful. In not one of the descriptions does it state what modifications AMD has made to every AGESA firmware to handle the present points, which in all honesty, is a fairly large factor to omit. There aren’t any indications in any way on ASRock’s X670E Taichi BIOS web page as to what every firmware modifications, and with no public notes out there to customers, it is a case of “replace to this firmware, it is really useful.”

So what do we all know in regards to the modifications? Nicely, we all know the crucial change going from AGESA 1.0.0.5c to the 0.6 and 0.7 variations is a lockdown on SoC voltage to 1.30 V. Beforehand, on the ASRock X670E Taichi with 1.0.0.5c; we have been in a position to set the SoC voltage to 2.5 V, which might virtually definitely lead to frying our X3D chips like an egg.

Picture Credit score: Igor Wallossek, Igorslab.de

The opposite modifications coming with AGESA 1.0.0.6, in accordance with Igor Wallosek, the Editor-in-chief of Igorslab.de, AMD has additionally added two new PROCHOT entries that time on to combating overheating. PROCHOT basically means Processor Sizzling, and it’s a managed mechanism that’s designed to guard the processor from overheating. There are two implementations right here. The primary is the PROCHOT Management mechanism which is exactly what it says on the tin. When the CPU hits an outlined worth, the part sends a PROCHOT Management sign, and the CPU attracts much less energy to attempt to mitigate temperatures and scale back the chance of harm. 

The second mechanism is PROCHOT Deassertion Ramp Time, which dictates how lengthy a processor can ramp up the facility after the preliminary PROCHOT Management sign has been disabled. Basically, PROCHOT Deassertion Ramp is the time it takes for the processor to get again as much as regular parameters, and totally different variables, together with cooling, the aggressiveness of mentioned cooling, and common warmth dissipation high quality, can dictate this time. If the processor is inadequately cooled, this may end up in an extended deassertion ramp time, whereas extra aggressive warmth dissipation strategies ought to theoretically enable for a faster ramp-up time.

The Story So Far: Players Nexus Deep-Dive – The Ryzen 7000 CORE Basic Points

Earlier than the rollout of latest firmware, Steve Burke, the Editor-in-Chief of Players Nexus, and his staff investigated the issues in-depth, together with wanting on the unique fried {hardware} from Speedrookie. This contains a defective and bulged out Ryzen 7 7800X3D processor and his burnt ASUS ROG STRIX X670E E Gaming motherboard. As a substitute of RMA’ing the {hardware}, Steve Burke reached out to the person and provided to purchase the {hardware} from him, minimizing the RMA lead time and permitting Speedrookie to buy new {hardware}.

The 38:46 long video is an excellent watch, and we definitely suggest that customers watch this, particularly for these extra within the inside workings (or points) of the Ryzen 7000X3D and 7000 sequence processors. To summarize Steve’s findings, we took away the next factors:

• AMD Ryzen 7000X3D CPUs are shutting down too late to mitigate bodily injury.
• ASRock, GIGABYTE, and MSI have a 116°C thermal journey level, and ASUS has 106°C, however generally did not work as meant.
• The thermal cut-off for Ryzen 7000X3D is meant to be 106°C and 116°C for Ryzen 7000.
• AMD EXPO enabled on ASUS is 1.35V on SoC voltage up till BIOS 1202 (AGESA 1.0.0.6).
• ASUS’s SoC Voltage settings have been/are too excessive.
• The AGESA firmware rollout has been nothing in need of chaos at this level.
• AMD is providing RMA (paying delivery each methods) on killed CPUs, even when EXPO has been used (a minimum of within the US)
• No phrase on if motherboard distributors will honor the guarantee (on the time of writing)

Whereas Steve and his staff at Players Nexus have gone deep into uncovering the basis causes of the issue, one factor stays abundantly clear: the problem is not only one which pertains to SoC voltage. There has definitely been some confusion between AMD themselves and its motherboard companions in implementing the suitable failsafe to stop the CPU (and motherboard socket, for that matter) from burning into oblivion.

The opposite downside pertains to ASUS right here, with a extra aggressive implementation of its SoC voltages, which Players Nexus confirmed of their testing as operating too excessive. Earlier than the AGESA firmware (1.0.0.6) replace by means of BIOS model 1202, ASUS was overshooting SoC voltage by 0.05 V over AMD’s newly imposed SoC voltage restrict of 1.3 V.

Picture Credit score: Gamers Nexus

Soldering leads and connecting the motherboard to a digital multimeter, a 1.35 V SoC setting inside the ASUS firmware (and with EXPO enabled) resulted in an noticed 1.398 V from an SoC pad. This was sometimes even larger when probed on the choke, at an eye-watering 1.42 V. This basically poses an issue that ASUS’s firmware and the SoC rails themselves aren’t cohabiting effectively with one another. An extra 0.05 V on high of the really useful 1.30 V is rather a lot, to say the least, however including an additional 0.05 V on high of that can undoubtedly result in dielectric degradation and probably result in useless CPUs and burnt motherboard sockets.

Doing a little preliminary testing on the impact of SoC voltage on stability on the most recent AGESA 1.0.0.7 (BETA) firmware, our G.Talent DDR5-6000 package of DDR5 reminiscence (2 x 16 GB) on the ASRock X670E Taichi would robotically preset 1.30 V on the SoC when making use of the EXPO reminiscence profile. To elaborate, sadly, we tried 1.15 V, which was a no-go, and even 1.20 V was a no-go. We ultimately settled on 1.25 V on the SoC for this package and our Ryzen 9 7950X3D, and we discovered stability in memory-intensive benchmarks was strong.

Maybe one of many largest issues to return exterior of Players Nexus’s testing was that AMD is now providing RMA help for customers who’ve used EXPO reminiscence profiles, one thing which usually voids the guarantee on AMD’s processors. Whether or not or not different areas intend to honor these RMA requests hasn’t been confirmed, but it surely’s unlikely to be a problem.

Nonetheless, it is a good gesture for customers with broken CPUs from a problem that’s solely not their fault. Motherboard distributors, alternatively, function inside their insurance policies and parameters, and it could be trickier getting an RMA on a broken motherboard just because AMD would not management motherboard distributors’ RMA insurance policies. We’d hope in good religion that motherboard distributors will honor the guarantee in situations of those burnout points, however we can not verify if they’ll presently.

Our Testing: Methodology, Check Setup, and {Hardware}

To summarize the rationale for testing AMD’s AGESA firmware, we aren’t attempting to replicate burning our Ryzen 7000X3D samples – sufficient processors have already been sacrificed for science. For that matter, we definitely did not see or odor any smoke coming from our ASRock X670E Taichi throughout testing, so we’ll take that as a very good signal.

Our function for testing is to spotlight any variations or variations in parameters and power-related parts coming from AMD’s newest AGESA packages. This contains rails like SoC voltage and Bundle Energy Monitoring (PPT) output from the AM5 CPU socket. As AMD has dialed down what customers and motherboard distributors can apply with reference to SoC voltage to 1.30 V, it is value noting that each one of ASRock’s firmware we have examined on the X670E Taichi on this piece robotically units SoC voltage to 1.30 V.  Whereas we do not have the required instruments and tools to solder results in the motherboard to look at ‘bodily’ voltages, we’re counting on HWInfo’s reporting prowess, in addition to a number of temperatures.

We additionally did some in-house stability testing in opposition to the brand new SoC voltage limits, operating a recent batch of assessments on our Ryzen 9 7950X3D paired with a G.Talent DDR5-6000 (2 x 16 GB) reminiscence package with its AMD EXPO reminiscence profile enabled. We discovered that issues weren’t secure till we utilized 1.25 V on the SoC voltage inside the firmware. Hitting as much as 1.25V on the SoC, our package was rock strong, even in memory-intensive workloads and benchmarks.

That has been our focus, attempting to push the reminiscence as onerous as we will to make sure full stability. Numerous the fanfare surrounding the problem, on the entire, has been unfairly placed on AMD’s EXPO profiles as being one of many causes; it isn’t. We all know that CPU-intensive workloads will generate extra warmth, however that is not what we have been investigating. We’re in search of variations in present and energy between the totally different firmware variations to see if AMD (and ASRock) has made optimizations inside its framework to cut back these components, with present, or extra particularly over present and the built-in failsafes being bypassed, which is without doubt one of the key issues within the burnouts.

Our take a look at bench for our AGESA (AM5) replace testing is as follows:

For our selection of workloads, we’re counting on the Reminiscence Check Suite from Openbenchmarking.org through Phoronix to implement our memory-intensive workloads. Though a few of these workloads aren’t optimized and do not run on Home windows, we used the CacheBench benchmark, which makes use of a number of information sorts throughout learn, write, modify, and skim/write/modify mixed. As a part of the LLCbench low-level architectural characterization benchmark suite, CacheBench is designed to check reminiscence and cache bandwidth efficiency and depends on a compilation of C++ Toolchains and compilers.

Learn on for extra evaluation.