MSI Raider A18 HX A9WJG-091TR RAM and SSD Incompatibility Issues - Is a BIOS Update Available?

[mimarsina158b02db]

MSI Raider A18 HX A9WJG-091TR looks great on paper, but I've run into two serious roadblocks:

1) RAM.
a) My Mushkin 96 GB RAM (2x48 GB kit) at 5600 MHz doesn't work with this device at all. The same RAM works fine on the Norse edition of the MSI Titan 18 at 5600MHz.
b) I took the high speed RAM from the Norse Titan and plugged it into the Raider. It worked, but only at 3200 MHz (half of the 6400 MHz speed it gets you on the Norse Titan), instead of the expected 5600 MHz.

2) SSD.
a) My 8 TB Gen 4 NVMe SSDs did NOT work on this device at all. They were either not detected, or one of them was detected as a 0 TB (zero) drive. These are Oyen Digital brand.
b) My 8 TB Gen 3 NVMe SSDs did NOT BOTH work either. Only one was detected, and the other slot was reported as missing. These are Sabrent brand.
c) I was able to verify that both slots work by plugging in two 2 TB Samsung SSDs. Both were detected successfully.
d) Please note that the SSDs in 2a and 2b above work fine on all laptops I've tried them before (including the Norse Titan!).

Seeing as the Norse Titan has no issues whatsoever with hardware that chokes the Raider, do we have any BIOS updates available for the Raider to address these issues?

Specifically, it's pretty crazy that an actually faster CPU device (the Raider's 9955HX3D defeats the Norse Titan's 285HX CPU, at least when both are running with 5600 MHz RAM) cannot handle SSD and RAM upgrades that were purchased from 2023 or earlier.

Please advise, as I will be returning the device if these issues cannot be promptly resolved.

 

[mimarsina158b02db]

So I actually also happen to own the Raider A18 HX A7V, the previous generation device with the 7945HX3D CPU.

1) I upgraded the RAM on this one to 96 GB via the Mushkin solution, and bingo! It works just fine, 5200 MHz speed.

Now check this out: According to my benchmark run (InstallAware Pre-Compiled Runtimes tool, which compresses nearly 25 GB of app runtimes in parallel, requiring nearly 100 GB free RAM and gobs of parallel CPU power), the 7945HX3D CPU is FASTER than the newer generation 9955HX3D CPU!

How does this compute?

Here's the numbers, all tests with 96 GB RAM:
MSI Norse Titan, 96GB RAM, 285HX CPU, 6400 MHz RAM speed: 11.13 secs (best speed)
MSI Raider, 96GB RAM, 9955HX3D CPU, 3200 MHz RAM speed: 11.30 secs (best speed)
MSI Raider, 64GB RAM, 9955HX3D CPU, 5600 MHz RAM speed: 12.30 secs (best speed) - this is slower due to insufficient RAM during the benchmark
MSI Raider, 96GB RAM, 7945HX3D CPU, 5200 MHz RAM speed: 10.30 secs (best speed)

Is the previous generation REALLY faster?

Or is this due to the fact that the newer generation CPU was running at just too slow a RAM speed? If we actually had it running with 96GB RAM at 5600 MHz properly clocked, would this then have realized a benchmark speed of 9.30 secs (give or take), thus realizing a VERY substantial speed boost over the best of this generation's Intel?

Edit: Just to comment on observed CPU speeds (as reported by the Task Manager [ditto the RAM speeds]):
- The Intel never exceeds 4.8GHz, even when just a couple of CPU cores are pegged
- The 9955HX3D never exceeds 5.1GHz, ditto
- The 7945HX3D realized 5.3GHz in the same scenario consistently

So it might have more to do than with the RAM speeds involved - where the prior generation might really end up being faster!

I N S A N E .

2) Even after the latest BIOS updates, this Raider also fails to recognize my 8 TB SSDs. SO DISAPPOINTING.

I've decided to return the Raider (9955HX3D), seeing as it won't likely ever support my SSDs, and given how messed up its RAM support is.

But yes, would be great if someone could comment on the AMD CPU generation's performance mysteries here.

 

[mimarsina158b02db]

Updates:

a) 2x WD Black SN850X 8 TB SSDs arrived today, and worked in the system. Seems like an incompatibility with the Phison E18 controller on the Sabrent and Oyen Digital drives.
I guess, at least with Oyen Digital, you get good value. Sabrent are expensive, don't replace dead drives, and are otherwise so not worth it.

b) 2x Crucial/Micron 5600MHz 48 GB RAM sticks arrived today. These too, worked in the system.
I guess, again, that the Raider/AMD HX3D series are just WAAAY TOO PICKY about their SSD and RAM components, then?

I'll soon be benchmarking the system - can't wait to see how it compares with the 7945HX3D (and, of course, the 285HX).

 

[mimarsina158b02db]

As shocking as it is, with all the upgrades, the build time has again peaked at 10.30 secs (give or take) - which means there really hasn't been *any* improvement between the 7945HX3D and the 9955HX3D - at least insofar as the InstallAware Pre-Compiled Runtimes benchmark is concerned.

Both chips would seem to defeat the latest and greatest offering from Intel, and although I'm curious how the 14900HX Intel CPU would fare in this scenario with 32 hyperthreading-included cores; unfortunately the atelier does not have any of those devices lying around for a benchmark comparison.

It seems to me some power savings in the 9955HX3D might account for this oddity in performance gains (or lack thereof), seeing as the Task Manager CPU frequency indeed never really remained high after the benchmark was left processing jobs on only 2-4 cores (my original findings above remain unchanged in this regard).

Well, I suppose it's been fun talking to myself on this thread.

 

[electrostingz]

Hello,

I've been reading your posts

The CPU's are boosting based on an algorithm that involves Power, heat and current draw of the CPU / VRM components. So if one is running hotter it may be limiting the boost or if the power consumption is higher for whatever reason / it's hitting a limit it would also run lower boost clocks. You would have to look at these values to determine why it didn't boost to the same level and how the individual cores were utilised during your test.

For instance using software see what the power draw / TDP is on the 9955HX3D when under this load, do the same for the 7945HX3D and monitor all cores boost frequencies. And do you know what the max power limits are for each one (CPU?) AMD showing 55-75W configurable, MSI might have set 85W on he 9955HX3D, what is the 7945HX3D laptop set to?

Ram throughput / bandwidth and latency are 2 other important factors, for a fair test I would consider trying to run the same spec ram on both machines to maintain the overall RAM latency and power usage (possibly some implications on the boost if the higher speed ram caused the CPU to work harder so it couldn't maintain the same boost).

As to the 14900HX, it's very similar to the 285HX, (EDIT oops, HT error) The 14900HX is only HT on the 8 Performance -cores, the 16 E-cores differ slightly and if your run time app is using the E-cores (most general use / gaming and software don't) there might be some advantage with the 285HX over the 14900HX, 4.6GHz vs 4.1GHz E-core boost but that depends what they can maintain underload and the power limits. Did the runtime software use the E-cores in your 285HX testing?

 

[mimarsina158b02db]

Oh nice to hear!

The 285HX doesn't have HT at all, and the raw performance clocks on the 14900HX seem much higher than they are on the 285HX. I wouldn't be surprised if it outdoes the 285HX in this benchmark at least.

The benchmark hammers all cores - it builds 199 application runtimes simultaneously, which consumes a lot of RAM, CPU, and HDD power combined. As the builds complete, you're left with just 2-3 large runtimes still building; which actually takes longer than all of the previous runtimes combined. In this sense, it's a nice benchmark that combines single core, multi core, SSD, and RAM testing all rolled up into one.

You can download it for free and run it on your own system from InstallAware; the "Pre-Compiled Runtimes Tool" never expires or requires a license. It'll auto configure itself to optimize builds based on the number of CPU cores you have available. I always test on the highest compression grade setting.

As it stands, the 7945HX3D is performing at the same level as the 9955HX3D with RAM that is clocked 400 MHz slower (latency might improve as the RAM is clocked down).

Since both devices are MSI 18" Raider's, with the same casing even. and have furthermore exhibited down to the same SSD incompatibilities as one another; I'd be hard pressed to consider the test unfair by any metric. If anything, the test is unfairly skewed in favor of the 9955HX3D, which I tested with the much faster Western Digital SSDs (and the faster RAM, too).

As such, it seems PC performance has really peaked over the last couple generations - with Intel outright regressing on the number of logical cores at least, and AMD barely holding it's ground due to buggy implementations of this or that?

 

[drek4]

Oh nice to hear!

The 285HX doesn't have HT at all, and the raw performance clocks on the 14900HX seem much higher than they are on the 285HX. I wouldn't be surprised if it outdoes the 285HX in this benchmark at least.

The 2XX HX series was meant exclusively for laptops, especially thinner ones, they just keep the 149XX HX around for those that are reticent to change.
The 2XX HX series has also the possibility of using faster onboard RAM than the conventional slotted RAM.

The benchmark hammers all cores - it builds 199 application runtimes simultaneously, which consumes a lot of RAM, CPU, and HDD power combined. As the builds complete, you're left with just 2-3 large runtimes still building; which actually takes longer than all of the previous runtimes combined. In this sense, it's a nice benchmark that combines single core, multi core, SSD, and RAM testing all rolled up into one.

You can download it for free and run it on your own system from InstallAware; the "Pre-Compiled Runtimes Tool" never expires or requires a license. It'll auto configure itself to optimize builds based on the number of CPU cores you have available. I always test on the highest compression grade setting.

As it stands, the 7945HX3D is performing at the same level as the 9955HX3D with RAM that is clocked 400 MHz slower (latency might improve as the RAM is clocked down).

Since both devices are MSI 18" Raider's, with the same casing even. and have furthermore exhibited down to the same SSD incompatibilities as one another; I'd be hard pressed to consider the test unfair by any metric. If anything, the test is unfairly skewed in favor of the 9955HX3D, which I tested with the much faster Western Digital SSDs (and the faster RAM, too).

As such, it seems PC performance has really peaked over the last couple generations - with Intel outright regressing on the number of logical cores at least, and AMD barely holding it's ground due to buggy implementations of this or that?

I think it's more an issue of heat, as we can see the desktop variants use much more power, thus increasing performance, so the Achilles heal of laptops is heat.
Most people want the laptops thinner and more portable, even the Titan and Raider are much thinner than back a few years ago, until they find more efficient CPUs,
it will be difficult to exceed performance expectations, but the 2XX HX series is an attempt at this, even if far from perfect in any way, is still a small step further.

As for AMD, what I prefer so much about the laptops I purchase them inside, is the fact that at much lower wattage, you have much more performance than the Intel variant,
but AMD and Intel will still struggle for the most powerful CPU, that is a good thing, but AMD wins hands down with battery performance, very important for me,
at least for my laptops I travel with, I love my AMD laptops for travel and love my Intel laptops at home.

 

[electrostingz]

I'll see if I can give the tool a go at some point and as to the 285HX not having HT , What was I thinking, I know this, got a little clumsy there.

 

[mimarsina158b02db]

The 2XX HX series was meant exclusively for laptops, especially thinner ones, they just keep the 149XX HX around for those that are reticent to change.
The 2XX HX series has also the possibility of using faster onboard RAM than the conventional slotted RAM.

I think it's more an issue of heat, as we can see the desktop variants use much more power, thus increasing performance, so the Achilles heal of laptops is heat.
Most people want the laptops thinner and more portable, even the Titan and Raider are much thinner than back a few years ago, until they find more efficient CPUs,
it will be difficult to exceed performance expectations, but the 2XX HX series is an attempt at this, even if far from perfect in any way, is still a small step further.

As for AMD, what I prefer so much about the laptops I purchase them inside, is the fact that at much lower wattage, you have much more performance than the Intel variant,
but AMD and Intel will still struggle for the most powerful CPU, that is a good thing, but AMD wins hands down with battery performance, very important for me,
at least for my laptops I travel with, I love my AMD laptops for travel and love my Intel laptops at home.

Well on the note of noise and heat, I can say the 9955HX3D is running a LOT cooler and quieter than the 7945HX3D.

However, I would have liked to see an option to, at least, put it on maximum throttle - albeit heat and noise. This basically informs my bias that 9955HX3D improves thermals and power savings only, at the expense of raw performance - which remains unchanged in real world use.

 

[mimarsina158b02db]

I'll see if I can give the tool a go at some point and as to the 285HX not having HT , What was I thinking, I know this, got a little clumsy there.

That'd be great, I'd love to see if anyone can realize sub-10-minute benchmark completions with the tool.

The closest I've come so far is in the 10:30 range with the AMD 9955HX3D (quietly) / 7945HX3D (loudly) - where the best of Intel remains stuck at the 11 minute mark.

Edit: Although, again, I'd love it if someone with at least 96 GB 5600 MHz RAM and a 14900HX Intel CPU (does that support 96 GB at 5600 MHz? I remember the 13980 HX supporting 5600 MHz at only the measly 32 GB capacity, and seem to think the 14900HX could handle it at only the 64 GB capacity) could run the benchmark, and let us know what they find in terms of raw speed there.

I wouldn't be surprised if that CPU ends up beating both Intel's latest offering and AMD both!

Please note that when running, I've disabled both Core Isolation (just in case) and Windows Defender (this one causes slowdowns of at least a minute, if not more, during testing). And always setting power to maximum performance.

 

[drek4]

Well on the note of noise and heat, I can say the 9955HX3D is running a LOT cooler and quieter than the 7945HX3D.

However, I would have liked to see an option to, at least, put it on maximum throttle - albeit heat and noise. This basically informs my bias that 9955HX3D improves thermals and power savings only, at the expense of raw performance - which remains unchanged in real world use.

Well then I consider this as a step in the right direction, since AMD has again surpassed Intel in more efficient CPUs, that's why AMD is my go to,
for my travel laptops, since being more efficient at lower speeds and thus cooler, is of upmost importance with thinner and portable laptops,
especially the 14'' and sub 14'' categories.

 

[mimarsina158b02db]

On that note, I'd have to disappoint you - erm, rather AMD.

On idle, this laptop, sadly, does emit a low pitched whirring noise; the 285HX is 100% quiet.

I hate it when they do that while idle.

 

[joao.sousa.domingo158102d7]

The Titan (285HX) is always quiet - I find it actually a well-cooled laptop (cpu helps). Battery is a misery indeed - I haven't left it reach 0, but I don't know if even a hour (on MSIhybrid). I'm sure igpu only does better. But c'mon.. the whole laptop is about performance, not mobility.

@mimarsina158b02db do you have the password for InstallAware? would like to run that Pre-Compiled Runtimes benchmark - whatever that is

 

[mimarsina158b02db]

The Titan (285HX) is always quiet - I find it actually a well-cooled laptop (cpu helps). Battery is a misery indeed - I haven't left it reach 0, but I don't know if even a hour (on MSIhybrid). I'm sure igpu only does better. But c'mon.. the whole laptop is about performance, not mobility.

@mimarsina158b02db do you have the password for InstallAware? would like to run that Pre-Compiled Runtimes benchmark - whatever that is

Oh yeah, that'd be great!

Just enter your email addy on their public download page, they regularly swap out passwords for new ones.

What's your rig? 14900HX maybe

Yeah, on laptops like these, the battery is best thought of as an uninterruptible power supply, isn't it

 

[joao.sousa.domingo158102d7]

Oh yeah, that'd be great!

Just enter your email addy on their public download page, they regularly swap out passwords for new ones.

What's your rig? 14900HX maybe

Yeah, on laptops like these, the battery is best thought of as an uninterruptible power supply, isn't it

First run:

That's with win OS SSD on a gen4 slot and laptop without battery - but I believe that doesn't affect these results.
I'll do more runs laters.

I'm latest Titan 18 HX AI A2XWJG (285HX) - oh and Kingston 128Gb 6400 (52-51-51-102)

 

[joao.sousa.domingo158102d7]

A few more runs:

I'll try again Sunday with new OS in raid0

 

[electrostingz]

I don't have enough RAM to be competitive, 32GB installed 5600M/T but it's a 14900HX laptop and I'm using modified CPU settings with HT disabled.

Just a note on that screenshot, it doesn't sustain 97 Deg C, only peaks momentarily at times so it's looks worse than it is.

 

[joao.sousa.domingo158102d7]

The only thing that is bothering me is the CPU as well but in Time Spy. I was able to get this after the latest bios update (.113):

But since then my CPU is only around 178xx and I can't go above that:

GPU is fine and always above 25400. It's only the CPU that is bugging me.. (and it's not because of updated RAM - it's exactly the same as it was before)

 

[electrostingz]

Time spy uses the last test (CPU test) to calculate this score so it comes down to max boost on all the cores, P and E when its running, did anything change with your boost power limits since running this test or temperatures higher?

 

[joao.sousa.domingo158102d7]

Time spy uses the last test (CPU test) to calculate this score so it comes down to max boost on all the cores, P and E when its running, did anything change with your boost power limits since running this test or temperatures higher?

yeah I know, only the third test matters. When I did the +24k final score above I never actually had HWiNFO opened.. So I don't know what sensors looked like then.
I never changed anything on CPU - never oced or undervolted on this laptop as I need it stable for my work - where do you change boost power limits? Intel XTU? I think mine is blocked as always (I believe would need to change advanced settings in BIOS first)

I'm attaching more recent runs of Time Spy - don't know if anything stands out.
(HWiNFO opened and screenshotted with start and end of Time Spy test respectively)