GIGABYTE AORUS AX370-Gaming 5 Motherboard Review @ [H]

FrgMstr

Just Plain Mean
Staff member
Joined
May 18, 1997
Messages
55,532
GIGABYTE AORUS AX370-Gaming 5 Motherboard Review - At long last we've finally got a full retail AM4 motherboard on hand to put through its paces. GIGABYTE's AX370-Gaming 5 is probably one of the best bets on an AM4 system right now. Unfortunately, that path will have a few bumps in the road. We fill in all the gaps and let you know what to look for.
 
  • Like
Reactions: Nobu
like this
Great and stable board for me during the short time period I had it. Updated to the Gaming K7 shortly after. Both very stable boards.
 
Thanks for the review ... good job as always.

I'd sure like to see the results with a Samsung 960 Pro NVMe M.2 with the latest NVMe version.
 
Should've got a k5! :p
I already got my motherboard, but a review of the b350 k* boards would be interesting, imo.
 
I was disappointed by the lack of options in the BIOS for overclocking on my K7. But to be honest I didn't even need anything fancy. Super solid boards.
 
I like the candid review, even though somewhat different between Kyle and Dan I can also understand why. There are some substantial variations between users and Ryzen. I have two 1700x, one does 4ghz with ease the other one (before I shocked it with 1.8v which now is more limited than before) had an issue doing even 3.8ghz (a lot of work to just get it there). Some have cold boot issues while others (like me) don't even understand really what the hell does that mean. Some get to DDR4 3200 with a few easy settings, others spend days trying to get there with the same ram, mobo, cpu models. So really this review or how it was put together reflects almost exactly what others are getting from RyZen. One may have a smile while the other is pulling their hair out.

It is almost too early yet to even do motherboard reviews due to the rapid progressions and changes going on. So I hope to see the older review motherboards updated with newer reviews as time goes on. Maybe not all at once but cycling through some of the older test boards to reflect any significant changes as the new board being tested is done.

Once things settle more, probably more after the May AMD AGESA microcode are in the bios, a detailed OC test between the top boards are done. What options the boards have compared to each other and how well they work or not. I don't think everyone will get it down as fast as the others.

Also can the B350 boards OC as well as the X370s? So much work or is it fun to be had here indeed :)
 
Thanks for the review ... good job as always.

I'd sure like to see the results with a Samsung 960 Pro NVMe M.2 with the latest NVMe version.

Unfortunately, I don't have that exact drive on hand for testing. Sorry. Still, it's interesting to me that a single M.2 drive is faster on AM4 than it is on Z270. We knew there was a potential for this when AMD gave us the initial chipset / CPU specifications. Unfortunately, X370 can't do two of them in RAID as far as I know. Any attempt at RAID would require running the secondary M.2 slot off of the chipset and that's likely to be as slow or slower than Intel's DMI 3.0. Alternatively, a motherboard manufacturer could split the dedicated x4 PCIe link for M.2 into two x2 PCIe links for M.2 based SSD's. Of course you are going to take a big performance hit from doing that.
 
I like the candid review, even though somewhat different between Kyle and Dan I can also understand why. There are some substantial variations between users and Ryzen. I have two 1700x, one does 4ghz with ease the other one (before I shocked it with 1.8v which now is more limited than before) had an issue doing even 3.8ghz (a lot of work to just get it there). Some have cold boot issues while others (like me) don't even understand really what the hell does that mean. Some get to DDR4 3200 with a few easy settings, others spend days trying to get there with the same ram, mobo, cpu models. So really this review or how it was put together reflects almost exactly what others are getting from RyZen. One may have a smile while the other is pulling their hair out.

It is almost too early yet to even do motherboard reviews due to the rapid progressions and changes going on. So I hope to see the older review motherboards updated with newer reviews as time goes on. Maybe not all at once but cycling through some of the older test boards to reflect any significant changes as the new board being tested is done.

Once things settle more, probably more after the May AMD AGESA microcode are in the bios, a detailed OC test between the top boards are done. What options the boards have compared to each other and how well they work or not. I don't think everyone will get it down as fast as the others.

Running motherboards past two different testing setups, environments and people often yields interesting differences. That said, Kyle and I seem to be on the same page with similar experiences most of the time. On occasion, we have vastly different experiences. One thing about this particular case is that Kyle and I didn't have all the same RAM to test with. We rarely ever do, but I never had anything on hand that behaved well with this motherboard and he did.
 
I experienced CLOCK_WATCHDOG_TIMEOUT BSODS upon my original windows 10 install on my Ryzen until I updated the USB drivers from AMDs website. (using USB WLAN at the moment). I didn't have any issues completing the install, but I am using NVMe unlike both Dan and Kyle.


The NVMe performance has been stellar, and the latest AGESA actually gave me 100 more MB/s than I was getting beforehand.

Honestly, if [H] had simply pre-ordered the boards before the launch from Newegg, they could have had boards in hand during the 1st week in March. The biggest knock I have is that there is absolutely no excuse for this review being this late.
 
Running motherboards past two different testing setups, environments and people often yields interesting differences. That said, Kyle and I seem to be on the same page with similar experiences most of the time. On occasion, we have vastly different experiences. One thing about this particular case is that Kyle and I didn't have all the same RAM to test with. We rarely ever do, but I never had anything on hand that behaved well with this motherboard and he did.
Which points exactly what many are experiencing with this platform, even if you had the same ram you may not get the same results. So the separate testing, use building etc. on the review I think brought a clearer view on what one may experience. In my case I initially bought the same Ram that reviewers were given, Corsair DDR 3000 ram, it was OK but only ran at 2666mhz - no matter what I did. Traded it in for GSkill Samsung B die stuff and DDR 4 3200 was easy and stable. Reviewers were for the most part getting DDR 4 2933 and a few were not. Anyways I liked how this review flowed and I too think Gigabyte did a damn good job getting that board as stable as it was. Compared to ASUS initial run with multiple people bricking their boards Gigabyte did really good here and helped make RyZen more legitimate.
 
I experienced CLOCK_WATCHDOG_TIMEOUT BSODS upon my original windows 10 install on my Ryzen until I updated the USB drivers from AMDs website. (using USB WLAN at the moment). I didn't have any issues completing the install, but I am using NVMe unlike both Dan and Kyle.


The NVMe performance has been stellar, and the latest AGESA actually gave me 100 more MB/s than I was getting beforehand.

Honestly, if [H] had simply pre-ordered the boards before the launch from Newegg, they could have had boards in hand during the 1st week in March. The biggest knock I have is that there is absolutely no excuse for this review being this late.
I don't know about that - HardOCP pretty much tells it how it is - that would have been some bloody reviews which in a span of time would paint a picture not too accurate. I think HardOCP holding off gave a more pertinent oversight for this motherboard.
 
Note: I did use an NVMe drive for the OS in this case because I had no choice. There were no issues doing that. Normally i use a SATA based SSD, which was problematic in this case. Later on I used a PCIe Intel SSD 750 for the OS so I could go back and do the proper NVMe testing.
 
I have the Gigabyte Gaming 5 AX370 and was wondering how to enter extreme calibration in the bios ? The manual said nothing about it and the enter key doesn't gave us different options. You have to enter a key to get the option and i doesn't know which one has i tried different key and gave me the standard option and nothing elese. What you did to have the extreme option in the bios ? The board work ok with the CPU at 3.8Ghz and the gskill flare x at 2400 16g (2x8g).
 
I have the Gigabyte Gaming 5 AX370 and was wondering how to enter extreme calibration in the bios ? The manual said nothing about it and the enter key doesn't gave us different options. You have to enter a key to get the option and i doesn't know which one has i tried different key and gave me the standard option and nothing elese. What you did to have the extreme option in the bios ? The board work ok with the CPU at 3.8Ghz and the gskill flare x at 2400 16g (2x8g).

Use the plus or minus keys with the load-line calibration option selected.
 
This review, while motherboard specific, speaks volumes about platform maturity. As was pointed out, it's nice to have something different to futz with for a change. This whole launch reminds exactly of the Athlon 64 X2 launch. Half the RAM on the market wouldn't work and what did had to go in slots X and Y with manual timing controls and we were all running around hacking BIOS and getting new official ones every week praying it fixed what the hell went wrong last week so we couold cap a new WR on whatever cooling pot we had manufactured this week.

I, like others see this platform as a value line up in the sense it could give me more value and socket longevity over Intel's offerings. An underclocked/undervolted 15/600 on an ITX board sporting an underclocked RX 560/1050 for media and 1080 gaming should be energy efficient,quiet and sit there in it's case doing that until death (either mine or the units).

Good overall look at where things are. Thanks gents.
 
I am really hoping for more boards to show up but these might be the best we'll see for some time. I'm hoping the QVL list expands, it's quite sad that my 3200mhz corsair can't get past 2400 but the 3000 runs at 2933 no problem. I intended to buy the 3000 but it's still sad that the higher rated memory is giving me more problems then (essentially the same exact thing) something rated to run faster.
 
Good review! Been wondering when H was going to start reviewing X370 mobos.
 
Are this motherboard and the K7 edition the same except that BCLK modifications are allowed on the K7? The K7 seems to be a really good price and I'm still spec'ing out new workstations. Thanks for the review it really helps purchasing decisions.
 
Are this motherboard and the K7 edition the same except that BCLK modifications are allowed on the K7? The K7 seems to be a really good price and I'm still spec'ing out new workstations. Thanks for the review it really helps purchasing decisions.

I will most likely grab the K7 once the stock stabilizes and this is the only "big" difference I know of.
 
I've had nothing but problems with this board. Shut downs that result in pulling the CMOS battery, lastest F5 BIOS does not throttle to idle clocks while overclocked. (F4 does) NVME drive goes missing if machine is restart from a bad overclock causing a manual power cycle.

I am hoping most of these issues were due to me not using RAM on the QVL. I bought some corsair dominator from the QVL so I hope this solves the issues.
 
I've had nothing but problems with this board. Shut downs that result in pulling the CMOS battery, lastest F5 BIOS does not throttle to idle clocks while overclocked. (F4 does) NVME drive goes missing if machine is restart from a bad overclock causing a manual power cycle.

I am hoping most of these issues were due to me not using RAM on the QVL. I bought some corsair dominator from the QVL so I hope this solves the issues.
I would suggest that any time an AM4 board shuts down from a bad overclock, you may be required to clear to the CMOS to get everything "back right." This again is NOT a GBT thing, it is an AMD thing....
 
Yeah I found that confusing... apparently Gigabyte has Gaming 5 and Gaming K5 X370 boards...?
Yep, the k5 has a clock generator and maybe a few other different features, but fewer power phases than the k7.
 
I've had nothing but problems with this board. Shut downs that result in pulling the CMOS battery, lastest F5 BIOS does not throttle to idle clocks while overclocked. (F4 does) NVME drive goes missing if machine is restart from a bad overclock causing a manual power cycle.

I am hoping most of these issues were due to me not using RAM on the QVL. I bought some corsair dominator from the QVL so I hope this solves the issues.

NVMe drives going missing should be a temporary thing. After a single reboot it should show back up.
 
NVMe drives going missing should be a temporary thing. After a single reboot it should show back up.


Not reboot, but a power cycle, yes it does show again, thankfully.

My NVME runs very hot underneath the graphics card too.
 
My NVME runs very hot underneath the graphics card too.
Yeah. If that is a concern, you purchased the wrong motherboard. Lots of board makers are addressing this exact issue.
 
When you use the thermal pad (and properly remove the film) it's supposed to help but it's difficult to discern by how much. I've taken temps through the holes in the shield and hit the same spot on the other drive and it does seem to be slightly better, but I'm not totally convinced. I'll break out the thermal probes on one of these MSI's next time and see what's going on. It's not the most accurate tool out there but I think I can get a reasonably good comparison between one of the drives uncovered and one under the shield. It's one of those things that I've been meaning to do but there's always something else to do. I did do this on the Z270i STRIX board and found that their heat sink design was less than ideal.

It wouldn't be a test on this board specifically. The shield appears to be the same on all MSI motherboards so it should apply to virtually any or all MSI motherboards.
 
Heh, with all the AGESA issues, maybe it's ok that ITX boards are taking their time....
 
Heh, with all the AGESA issues, maybe it's ok that ITX boards are taking their time....

The ITX offerings are taking their time because the motherboard manufacturers put less priority on bringing them to market because they aren't high volume sellers.
 
My Ryzen experience has been with a 1700, Crosshair VI Hero, Samsung 960 EVO M.2, and G.SKILL TridentZ RGB 16GB (2 x 8GB) DDR4 3600 F4-3600C16D-16GTZR for DRAM. I have had a very different experience than what I read in this review. I did update the bios/UEFI right off the the bat to the latest before doing anything else. After that, I booted into UEFI, set some memory timings manually, rebooted, installing Windows 10 from USB, and bang everything was stable and just worked. From there I started working on memory, training incrementally to 3200MHz. A bit of tweaking, and I had 4GHz on the CPU (I now am usually running at 3.9 to 3.75Ghz with P-state just because it runs so cool at those speeds with lower volts.

I subsequently dropped in a second identical set of DRAM for a total of 32MB. My sound card hates BCLK increases so I am running at 2933MHz, though 3200MHz is doable if I am willing to go with motherboard audio. The memory options - and really all the UEFI settings are pretty deep on this board. The latest official UEFI version is 1002, but there are a bunch more test bios versions for those that want to try various iterations and the capabilities they have. There are issues for sure, like some limitations with VID with P-states, though I know these are being worked on.

Maybe I was just lucky and my component choices were fortuitous, but overall I think I'm glad I shelled out a bit more for the Crosshair VI. Like all AM4 adopters I'm hopeful about the May AMD update - maybe I can get the 32GB up to 3200 or better at 100 BCLK then.
 
From the article:
>> 4x PCIe lanes that can be used for NVMe devices or split between
>> additional SATA ports and a 2x PCIe NVMe M.2 implementation.

Regardless how AMD or your motherboard's manual might sell those
lanes, all four Ryzen NVMe lanes can be repurposed for PCIe 3.0 and
are GPU capable. An adaptor even exists (from NVMe to eGPU, if I
can recall where it was I put that darn photo.)

Proof is in ASROCK B350 motherboards that bypass B350 chipset
to give 4x 3.0 to to the second GPU slot. Use of NVMe 3.0 disables
the 2nd GPU slot. "**If M2_1 is occupied, PCIE4 will be disabled."

This is not the usual 4x 2.0 offered by chipset, but 4x 3.0 offered by
Ryzen SOC. Clearly those lanes are more flexible than your article
suggests.
 
Last edited:
From the article:
>> 4x PCIe lanes that can be used for NVMe devices or split between
>> additional SATA ports and a 2x PCIe NVMe M.2 implementation.

Regardless how AMD or your motherboard's manual might sell those
lanes, all four Ryzen NVMe lanes can be repurposed for PCIe 3.0 and
are GPU capable. An adaptor even exists (from NVMe to eGPU, if I
can recall where it was I put that darn photo.)

Proof is in ASROCK B350 motherboards that bypass B350 chipset
to give 4x 3.0 to to the second GPU slot. Use of NVMe 3.0 disables
the 2nd GPU slot. "**If M2_1 is occupied, PCIE4 will be disabled."

This is not the usual 4x 2.0 offered by chipset, but 4x 3.0 offered by
Ryzen SOC. Clearly those lanes are more flexible than your article
suggests.
Yes, they are general purpose, but on X370 board making a third x16 (x4 real) slot for the sake of 3-fire is just... pointless.
 
The M.2 drive location on the X370 Taichi is better so I went for that model instead.

I can certainly understand that.

M.2 is a stupid form factor designed primarily for mobile applications. I wish the SSD makers would get away from it and make more U.2 options. M.2 is especially problematic in cases where a motherboard will have more than one slot. M.2 eats up a lot of PCB real estate and often makes for a difficult "upgrade" installation down the line as you'll have to remove expansion cards to reach the slot. The tiny screw and standoff locations can be a pain in the ass to deal with as well. Lastly, M.2 is stupid because it generally forces manufacturers to put the slots underneath a video card or in a location where ambient temperatures aren't ideal.
 
Wow, when I read the first few pages of this review, Dan's negativity about the Ryzen platform lowered my mood !!! :(:(
I don't have the Gigabyte board, (Asus x370 prime), but I had no issues installing windows 10 on a Samsung SSD drive, but I still haven't been able to get my corsair DDR4 3200 memory above 2400. I am certain the platform/software will get better with time.
 
It isn't as if X370 is a bad platform, it just isn't quite up to par with Z270 and I was hoping AMD would do a little better on that front. As I stated in the review, I don't think the issues are huge and they shouldn't impact most people but Intel's Z270 is more flexible than AMD's X370 is. It's that simple.
 
>>M.2 is a stupid form factor

We will eventually see those moved to the back (Biostar already does this).
Case and motherboard makers need to agree where to put access holes.
Can't even get at backplate of CPU as it is now, is kind of rediculous...
 
Back
Top