Gigabyte Aorus X370 Gaming 5 Review

[Ket]

Gigabyte Aorus Gaming 5 Review
** EDIT: 13/ 09 / 17, All images are still from the original review, now the gentlemans agreement time is over I'm updating all of this review with my original writeup that I have updated slightly based on a quick revisit and look of the most recent UEFI for the board. Let it not be said that I don't give manufactureres a fair crack to sort themselves out, Gigabyte have had 6 months now.**

Introduction

Welcome dear readers to another no-nonsense review, this time we will be taking an in depth look at the Gigabyte Aorus Gaming 5 for the newly released Zen line of processors from AMD. The Aorus branding is a relatively new line from Gigabyte designed with the intention of going toe-to-toe with the best gaming and enthusiast products the industry has to offer. Gigabyte in more recent years have become more better known for their graphics cards rather than their mainboards but once upon a time Gigabyte made some of the best mainboards you could buy backed up with superb BIOS support, are Gigabyte returning to those lofty haydays? Let?s find out.


The X370

Gigabytes Aorus Gaming 5 packs AMDs (designed by Asmedia) X370 chipset, it?s the top of the line currently offered for a Zen based system. Let?s have a little look at what the X370 gives us;



This is what AMD calls high end. I don?t know about you but the X370 looks mid-range to me while the B350 looks low end. It?s disappointing to see the X370 chipset only providing 8x Gen2 PCI-E, why not Gen3, and why not more than just 8x? AMD could have used this opportunity to get one over intel here, but alas they have not. PCI-E handled by the CPU is a much more familiar arrangement however so nothing surprising there. Personally I think the X370 chipset looks a bit on the lacklustre side as is the inclusion of only 2x USB 3.1 Gen2 ports. At least the X370 has been given what?s known as ?Overclocking+? over the B350. Still, let?s not judge a chipset by the cover.

Let?s move on then.


The X370 Aorus Gaming 5

Now let?s get this show truly on the way. Fortunately it was nice outside earlier even if it was a little cool and windy. A cool breeze, an ominous sign perhaps? We?ll find out.


Oh, hello there chimp! What?s that? Do I want a cup of tea? I certainly do! Very kind of you to offer, could you just sit on that box for a moment first? Thanks.

So this is what you can expect the X370 Aorus Gaming 5 to come in, looks pretty nice doesn?t it? It feels nice too. I can?t say I?m a fan of just how little protection the board is offered, anti-static bag along with a thin, elevated cardboard shell which when lifted out reveals the included accessories and user manual with driver DVD, but top marks for presentation at least. As long as you?re chosen supplier packages everything well there shouldn?t be any problems.



Here we have the board itself, Phases look to be 6+4 which should be able to provide a good deal of power when Overclocking, the board is a very nice and clean design, and the board also has LED lights along with PCI-E and DIMM slot strengtheners. Has anyone ever physically broken a DIMM bank or PCI-E slot? No? I thought not. I wonder if there?s anything that has been sacrificed in order to accommodate these features. We will find out.

There are issues that need mentioning however, I don?t like the placement of the M.2 slot, no sir, not one bit. Why would you put one component that can possibly get quite toasty next to a GPU that will get very toasty?
I also don?t like that THREE fan headers are stuffed on the bottom right-hand corner of the board, that will make working with it on systems that use a good number of fans far more awkward than it needs to be, thanks to this poor fan header layout there?s going to be a lot of folks needlessly having to use extra splitters.

The Mosfet heatsinks are not using a heat pipe either, but the heat sinks are quite sturdy so maybe we will be ok here, we?ll find out later. I can?t say the design of the heatsinks from a functional perspective fills me with a particularly warm fuzzy feeling, despite them looking quite robust. The soldered EEPROM chips while looking neater than the socket variety I don?t like, you can?t replace either of those should one of them physically fail like you can the socket variety, but at least there?s a backup.

That shroud you see over the I/O ports obscuring one of the Mosfet heatsinks is plastic. I don?t like it, not least because it is obstructing airflow over the VRM heatsink. Plastic in such abundance on something that is supposed to be a premium product just looks tacky, and yes it is a premium product, it uses the top tier X370 chipset for AM4. That makes the board top tier too.



The included accessories with the Aorus Gaming 5 are certainly nothing to get excited about the most notable thing is the inclusion of 2x sensors. The included SATA cables are also reduced to less than adequate, the board has 8x SATA ports and only included 4x SATA cables. I?ll be scoring on that. When you are getting sticker sheets and accessories like Velcro ties even the manufacturer must know it?s a weak accessories bundle. For 170 pounds, and that?s the lowest price I?ve seen with the average price being 200-230 pounds at etailers right now, the accessories are lacking. This light bundle is what I would expect to get with a much cheaper board, by which I mean one of the B350s you can pick up for 100 pounds or so. We now know what took a hit in order to add the LEDs and rather useless (albeit nice looking) DIMM and PCI-E slot strengtheners to the board.

First impressions are a mixed bag then, quality of the board appears good but with some missteps, and the accessories bundle leaves a lot to be desired with not even enough included SATA cables for all of the SATA ports the board has.


Laid to Bare

This section is where I pull everything off of the board to scrutinise it in detail, so let?s take a look at what keeps this board cool, and if there are any major issues.



Let's start with an overview, shall we? On a positive note the 6+4 phase design looks good, there's proper audio caps for the dual ALC1220 CODEC, the stock AMD mounting mechanism works for both clip springs and screw downs, albeit I wonder why AMD didn't design one mounting mechanism for all of their Wraith coolers. It would be easy enough to do and save them money in the long run. The board looks a lot better for removing those nasty cheap pieces of plastic too, and while I?m not a proponent of LED light?s on mainboards I?m sure they will look nice. My concerns for the Mosfet heatsinks are also partially alleviated, the heatsinks themselves look adequate but they could be far, far, better for little to no extra cost on Gigabytes part.



The first thing that sticks out like a sore thumb is that the piece of plastic covering the dual ALC1220 CODECs isn?t going to provide any EMI shielding to help guard against crosstalk from other components. Just what exactly is that bare piece of plastic supposed to shield against Gigabyte? If you are going to put something over the audio components make it worth the effort, use some EMI shielding. It?s cheap.

The next thing to stand out is the apparent contact the Mosfet heatsinks make, or lack thereof. Those Mosfets will cook themselves good under load without much better contact. I?ll be scoring on that, poor thermodynamics.

Finally let?s talk about the thermal solution itself for a moment. While I can't tell you the exact grade of the thermal pads I can tell you it will be "middling grade", because that's what all manufacturers use for stock components. This means that the pads probably have 3.2W/m-K properties, not particularly stellar, especially as far as those poor, starved of good contact, Mosfets go. The chipset likely won't fair too well either, not where the thermal pad is inadequate in that case, but because the heatsink itself is just inefficient. It is literally just a lump of metal with a hole though it's centre, no fins, nothing. It?s just a bad heatsink from a thermodynamics perspective.

It?s quite ironic that Gigabyte put that huge Aorus heatsink on when something smaller (thus cheaper for Gigabyte) following standard thermodynamic design, would perform better.

Now for some closer up images, because I know you all want it.








The VRMs are not the best quality on the AX370 Gaming 5 (or 7), they are PowIRstage, I could not find out any details about amperage but assuming they are on-par with NextFET IR3553s they will handle up to 240A. This makes high grade thermal pads and proper contact from the heatsinks essential in my opinion given the uncertainty in the exact quality of these VRMs particularly as I noted quite unstable CPU voltage. Every little helps.



Test Setup

With a new platform comes a new test setup.

CPU: AMD Zen 1700
Mainboard: Gigabyte X370 Aorus Gaming 5
RAM: 2x8GB G.Skill RipjawsV 3200MHz 15-15-15-35
GPU: 4GB GTX 980 @ 1.5GHz, 7.6GHz memory
Storage: 250GB Hynix SL301 SATA SSD, WD 120GB M.2 SATA SSD (OS drive), 2TB Seagate Barracuda
Opticals: 24x Lite-On iHAS324 DVD-RW, 16x HP BH40N Blu-Ray
Sound: Xonar DX 7.1, Dual Realtek ALC1220
PSU: EVGA 1000w Supernova G2
OS: Windows 10 Pro x64 (latest ISO) and all updates
Case: NZXT Phantom 530

Drivers Used

Chipset: v17.2.1
Intel LAN: v22.0.1
Killer LAN: v1.0.1028
GPU: GeForce ForceWare 378.72
Realtek Audio Driver: R2.81
Xonar DX Audio Driver: unixonar 1823
Asmedia USB3.1: v1.16.36.1

**IMPORTANT NOTE!!** Gigabyte only list the ancient 1.16.28.1 driver on their website, installing it caused a near instant BSOD. Go download the ASM1143 1.16.47.2 driver instead.

There we have it, a new beast for a new age. Let the tests begin.



Storage Performance: SATA HDD, SATA SSD, M.2

A quick on-point test now to check there?s nothing out of the ordinary going on.



So starting with the most boring device then we see the Seagate Barracuda 2TB 7200RPM mechanical drive performing within normal boundaries, nothing to really see here.



Having a peek now at a SK Hynix SL301 250GB SATA SSD we see some rather good numbers, once again nothing out of the ordinary to report.



Lastly the overview for the 120GB Western Digital Green M.2 drive, which is a rebadged SanDisk. I don?t see anything out of the ordinary here either, performance for a 120GB SSD seems right around the mark it should be.

Overall performance of storage devices seem to be slightly down but nothing I would consider worth being worried about. AM4 is a new platform that is bound to have some minor kinks to be ironed out still.


Benchmarks - SMT ON

This portion of the review has taken far longer than it should of, namely due to how fickle the Gaming 5s UEFI currently is. In an attempt to keep results as consistent as possible the CPU multiplier has been set to 37.25 so the Zen 1700 constantly runs at a maximum stock boost frequency of 3.7GHz, actual speed is a touch over that at 3717MHz.




First up then, AIDA64 shows reasonable results for Read, Write and Copy tests although I did think these results would be a little higher. On the whole I don?t think there?s anything to particularly worry about here though.




Latency seems right on the money for this point in time, no worries here.




It?s good to report that single and multi-thread performance puts in a strong showing from the CPU-Z benchmark. Zen is looking rather promising.




I was unsure if I wanted to include these results as I don?t like comparing a Quad CPU with an Octa-core CPU, they are in totally different leagues. In the end I decided they merited inclusion but as I don?t have at least a Hex intel equivalent CPU to hand the i5 3570k will have to do.

Geekbench once again shows Zen being an absolute monster of a CPU, that single thread score is almost as high as a 3570k clocked to 4.5GHz, and the Zen 1700 is doing that at only 3.7GHz, impressive.




Without question, the Zen 1700 simply wrecks the overclocked i5 3570k here, there?s really nothing else to comment about.




Using the Xeon X5650, a 12 core, 24 thread part, from Cinebench for comparison, it?s yet another strong showing for the Zen 1700 on the Aorus Gaming 5.



Ignore the latency bug results in the stalker graph the issue has since been resolved

Here we have our final overall system benchmark test, Stalker ? Call of Pripyat is based on X-Ray engine 1.6. The engine is notorious for the heavy load it puts not only on the GPU but CPU and memory as well making it an exceptional choice in identifying a weakness in any of these areas.

As we see in these results there is another strong showing. The sooner AMD and manufacturers fix memory latency shortcomings by adding more timing options the better as I can see some nice gains here if / when that is done.




Another more system orientated test now, these are once again some solid numbers but I had expected to see a little higher, there is still work to be done with the platform yet.




Moving on then let?s have a look at how performance is with a few physics tests, we see that Zen completely wipes the floor with the 3570k. Anyone with a similar system can be safe in the knowledge that there is a significant performance improvement to be had here.


The UEFI

The tricky bit here really is trying to find a few pages with anything interesting as the Gaming 5 UEFI is positively barren. There are not even basic options to disable on-board devices such as the audio or LAN. All screenshots have been captured from UEFI F5c. As of UEFI F9d very, very little has changed or been added


Overclocking options up first;











Gigabyte can?t be faulted for the lack of memory options at this stage it?s the same for every AM4 board. What must be of some embarrassment for Gigabyte though is the UEFI being barren of other OC features such as spread spectrum control which would add a bit of wiggle room to help make up a bit for the lack of a clock generator chip. Since AGESA 1.0.0.6 AMD have added many more memory options.

Now let?s have a look at the general BIOS and peripheral options;






More or less your standard fare here, although the question does have to be asked; just where are the options to turn off on-board devices such as audio and LAN? The answer to that is that there aren?t any. Not yet anyway. F9d has the option to enable or disable audio

Let?s take a look at the chipset and fan options;



Oh good lord, there aren?t even any Southbridge options for power saving features. What?s going on over there Gigabyte? ...and of F9d there still isn't any access to the chipset options.



Our final screen brings us to the options for fan profiles, well, at least we have more than a handful of options unlike everywhere else we have looked, it?s practically an all you can eat buffet in this menu.

It has to be said the current state of the AX370 Gaming 5 UEFI is nothing short of a monumental disappointment, and it looks untidy with all those duplicate entries that I?m told are ?by design?. Ok Gigabyte, I get it, you think people have the worlds shortest memory span. Personally I think people just want some actual useful options not to be told the same thing two or three times over in different menus, agreed? The UEFI distinctly has that look and feeling of being very rushed and poorly thought out. It?s also ugly and clunky to use.

What would be wrong with having options like USB support, PS2 device support, chipset power saving features, spread spectrum, spread spectrum options, HPET and USB power down options available for the user to adjust as needed? All of those options appear to be in the UEFI and functioning, their just hidden, I?ve checked with my own AMI tools Gigabyte. So why aren?t they available to the user? Don?t treat us like idiots.



Overclocking

As we approach the final part of this review we will take a look at the Overclocking abilities of the Gigabyte AX370 Gaming 5. Below is the result I was able to get with SMT ON, if Gigabyte are willing to put in the necessary time to mature the AX370 Gaming 5s UEFI I can see it being a capable overclocker, but time will tell on that front. I would try other DDR4 kits but the other kits sitting around are definitely not compatible with AM4, at least at this time.



No matter what voltage or combination of settings we tried with the stock F5c UEFI and the extremely limited options within it we could not pass 3.95GHz and be stable. The results themselves are also fairly unexpected, they are inconsistent. Current compatibility issues are more than likely the culprit here. Hopefully as the AGESA code matures it will help with overclocking to some extent in all areas. It was also noted that CPU voltage never remained all that stable, constantly bouncing between 1.32v and 1.39v with a set voltage of 1.35v.




With the modded F5c UEFI I made, 4.1GHz so far would appear to be stable. We got these speeds with nothing more than 1.4v, the stock AMD Wraith Spire running full speed and some Coollaboratory Liquid Ultra.

There are hints that 4.2GHz could be stable too but we need a better cooler for that, unfortunately none of the AIO liquid coolers sitting on the shelf have AM4 brackets available for them at this time.

It shouldn?t come as a surprise that I used the BIOS/UEFI tools at my disposal to fix up Gigabytes UEFI given the state it?s in right now but I won?t be releasing any modded UEFI for now, I?m going to give Gigabyte the chance to sort the UEFI out, we'll see how that goes. As of UEFI F9d, its still in just as bad of a shape as back in March, with the exception of one or two extra options and a lot more memory timing options. All work done by AMD in the Agesa code.



Conclusion

As always I will be objective about my conclusion, basing it on available data. As such this makes it a tricky one, so this conclusion is going to be a little long. I?ll start with the AX370 Gaming 5 itself from a hardware perspective. From this angle I do genuinely really like the board, it does have a lot going for it even the LED lights, something I generally do not favour as I always worry more time has been spent on looks rather than raw functionality, I enjoyed. Minor missteps aside such as that nasty plastic shroud over the VRMs, no EMI shielding over the dual ALC1220 audio CODECs that Gigabyte seem so proud of, and what I would consider questionable VRM quality and not really quite enough of them (it should be 6+6 in my opinion) the physical side of the board is really quite good. The board also did not brick itself like the Asus C6H (test board I had died after around 5 POSTs), so that?s something.

Where the AX370 Gaming 5 is phenomenally let down though is with the UEFI, the most feature rich part of the UEFI, and I would hope is a sign of things to come, are all the options available for tailoring fan speeds. Aside from that the UEFI is stripped down and devoid of almost everything, the layout is pretty clunky, there are duplicate entries in menus ?by design? that I simply cannot explain, and apparently redundant options for various settings, or at the very least options that need renaming to put their function in to some sort of perspective. I don't feel it's fair to comment on the overclocking capability of the board due to the limited UEFI options at this time, along with early teething problem encountered either. While all manufacturers are having issues with their AM4 boards right now they all at least have a far more feature rich UEFI along with a considerably less clunky appearance.

So that brings us to price, at time of review some googling reveals in the UK this board, as I mentioned earlier, retails for anywhere between ?170 to ?230, the cheapest place being AWD-IT, props to them for the board for testing. So if you wanted to take your chances and hope the UEFI significantly improves that would be the place to buy from. I was genuinely enthusiastic about testing the AX370 Gaming 5 but the nigh-empty UEFI sours the whole experience considerably. At ?170 the board is placed right in the middle price-wise in terms of the X370 boards you can buy and should you choose to buy from whoever your tried and trusted etailer may be instead of AWD-IT you are likely going to be looking at ?200 or more, this board is meant to be high end and has a price to match that. However at this time I can?t in good conscience recommend you all rush out and buy this board. The many small shortcomings, and the big one of the severely under developed UEFI that looks like it hasn?t even made it out of alpha testing stages yet make it a board to keep a eye on and monitor it?s UEFI development at best.

In closing, I will address this to every manufacturer out there producing, or going to produce, AM4 mainboards; You knew when Zen was going to arrive, you surely knew a lot of people were going to be interested in Zen, you should of dedicated more resources to the development of the new AM4 boards and taken initiative to ensure the boards would be in a far more retail ready state than they are. You did none of that, instead putting (probably) almost all of your resources in to the Z270 boards. Do not leave the AMD consumers high and dry. Now do your jobs, step up, and sort your respective messes out with your AM4 boards.


Hardware Functionality & Quality, Accessories: 23 / 30
Board Aesthetics: 10 / 10
UEFI Functionality & Quality: 10 / 30
Performance: 24 / 30

Final Score: 67%


As is fair of any revisit the score has been amended to reflect the latest state of the board.

Hardware Functionality & Quality, Accessories: 23 / 30
Board Aesthetics: 10 / 10
UEFI Functionality & Quality: 14 / 30
Performance: 26 / 30

Final Score: 73%

[StyM]

HB bridge included ?

[Ket]

The only bridge included is the one you see in the picture, which is a GC-SLI2P. It's not a bendy bridge either, old style hard PCB variety. Results should hopefully follow soon making sure all hardware drivers are the latest, working out a kink or two, updating Windows itself, etc.

[Ket]

Bugs: UEFI F5C


Bugs Found

1. Mouse input / responsiveness is laggy in UEFI. Experienced with hardware: Roccat Kova+
2. Resolution appears to be stretched at POST and with the Windows 10 logo while booting when Aorus splash screen is disabled. Experienced with hardware: BenQ GL2450 connected via HDMI, Gigabyte Windforce3 GTX 980
3. LED keyboards are unusable in the UEFI once they enter sleep mode (keyboard backlight goes out) until a reboot. Experienced with hardware: Ares K5, Sunt PK35
4. XMP profile must be loaded before memory multipliers will work. Experienced with hardware: G.Skill Ripjaws V 3200MHz (F4-3200C15D-16GVK)
5. Memory multiplier is limited to a maximum of loaded XMP profile. There must be a way to be able to set higher multipliers.
6. XMP and memory multiplier options are in both Advanced Frequency and Memory Settings menus
7. CPU Clock Ratio is in both MIT and Advanced CPU Core settings menus
8. Apparently redundant naming for some settings in the UEFI. Eg; CPU LLC: Normal, Standard, Medium. Three settings, all essentially with the same meaning.
9. Extremely bad latency, possibly Agesa related and Windows 10 related.

[chew*]

Should probably note besides sensors that the RGB cable is included, clearly see it in pic, not sure if your aware that that is what its is. While that is not much interest to me the case modding crowd etc etc might care.

Latest F5c bios is linked in my thread. probably should test it with that since public now.

[Ket]

Yeah I know, but christmas lights on a mainboard isn't something I consider to be even remotely beneficial or of any interest what so ever, their just lights. Thats why the cable is pictured but not mentioned. I'll be sure to grab F5C, F3 definitely has it's share of quirks in the UEFI itself.

[chew*]

I posted a workaround for SMT disabled testing with all 8 cores with 1700 at desired frequency if you wanted to test that.

The issue does not effect X models and may be widespread across all vendors.

waiting for confirmation from another qualified source on a different vendors product.

[Ket]

Updated main article, test results next.

I posted a workaround for SMT disabled testing with all 8 cores with 1700 at desired frequency if you wanted to test that.

The issue does not effect X models and may be widespread across all vendors.

waiting for confirmation from another qualified source on a different vendors product.

I'll check the workaround out.

[Ket]

Updated post #4 with initial F5C UEFI test results, in short it's worse than F3.

[chew*]

Solution for bclock, hardmod a clock gen chip, there wont be a solution for g5. lacks turbo bclock feature of g7 in feature list on site.

[Ket]

Solution for bclock, hardmod a clock gen chip, there wont be a solution for g5. lacks turbo bclock feature of g7 in feature list on site.

The G5 is looking more and more like a budget board with a premium price tag at this rate... I'll reserve judgement until I finish the review though, perhaps Gigabyte still have features and stability improvements they will push out before the review is complete. Also found a few more issues, added to post #4.

Hopefully I can get the rest of this review done now.

[Ket]

Updated article with benchmark results.

[Ket]

Updated article, almost done now one final editorial pass and a few charts is all that remains. Article is 18 pages long, for anyone curious.

[cdawall]

So it has 3.0 for PCI-e not 2.0 it literally says that in the AMD graph you took a screenshot of.

The limitations of the chipset for this mainstream board you listed as a complaint don't make any sense, the roughly follow the same specifications of intel with Z170/270.

No board in that price range or even into the $400 range for intel has a heatpipe based cooler for the MOSFET's, the plastic is also not a fire hazard. Come on really?

Also just as an FYI you can screenshot in the BIOS now...LOL

[Ket]

So it has 3.0 for PCI-e not 2.0 it literally says that in the AMD graph you took a screenshot of.

The limitations of the chipset for this mainstream board you listed as a complaint don't make any sense, the roughly follow the same specifications of intel with Z170/270.

No board in that price range or even into the $400 range for intel has a heatpipe based cooler for the MOSFET's, the plastic is also not a fire hazard. Come on really?

Also just as an FYI you can screenshot in the BIOS now...LOL

- I like to comment on things, even the obvious because I got tired of people asking why I didn't comment on things more
- Yes they do CLICK
- I always consider the worst case scenario, a case with poor airflow, combined with a board that might have a few years on it but a owner that doesn't know any better to replace TIM / thermal pads. I'd bet on that piece of plastic starting to melt in scenarios like that especially if the mosfets are getting toasty, who knows what may happen from there. Plus I just don't like seeing cheap plastic.
- Since when does the X370 Gaming 5 UEFI support screen captures? I looked everywhere in F5C and couldn't find any reference for screen captures, I hit the typical hotkeys while in the UEFI with a USB stick in and still nothing then either lol.
- Midrange in the UK is around the 140 pounds mark, the Aorus Gaming 5 in the UK retails on average for around the 200 pounds mark. In the UK that puts the Aorus G5 firmly in the upper tiers. Only AWD-IT were selling the Aorus G5 for 171 pounds, but they have since stopped listing the board.

[cdawall]

So you commented on something that wasn't true and then ignored. Ryzen supports 3.0 pcie.

Z270 using the chipset to "create" more pcie lanes doesn't make more lanes. There is still only a 4.0 link to the cpu and 16 actual lanes.

I have never seen that piece of plastic melted on any board ever. I have even had boards that the mosfets have run so hot they have popped. Also remember this is a 95w TDP max board that has the ability to throttle the vrm. I also notice you didn't comment on what the vrm rating was. At all.

Never seen a new GB board that doesn't allow screen capture...

It's still a midrange board hence why it is the M5 and not the M7. I could care less what overpriced amount you spent to buy an item during release month.

[NEOAethyr]

For bios caps, if you need to use a cam to doit, set the cam's exposure lv to the lowest it can go, in the neg's.
That will help with taking screens of the monitor.
It increases the shutter speed but also reduces the total light.
I also recommend totally not using flash, because it's showing up in the middle of your pics (monitor pics, the hardware pics are fine).

The plastic shielding thing...
I wish board makers were more serious about this, a meshed metal shield on the board would do fine, and allow heat to escape.

The chipset thing, pcie 3.0 and 2.0.
From what I understand, the cpu is 3.0 16x (actually I think it was 20x), and the chipset is 2.0.
Which is fine for most things, other then maybe m2 cards perhaps (even in that case, "in theory" it should not make a difference).


Your latency bug...
That's an odd one, I don't believe removing the mem should do anything.
However depending on how you boot up settings could make a difference.

Say or ex, you set 2133 mem, and it shuts down to reboot.
After that you set 2666 mem, and say it just resets, it doesn't power cycle, in this case there could be a prob.
Either the mem is gonna be hella fast, stuck with 2133 timings in 2666 mode, or it could go wrong.

My intel has an issue like this, at 100mhz bclk, 2400mhz div is ok, the 2666 div is not ok.
The 2666 has a mem write bandwidth bug.
However, if you manually power cycle it the bandwidth is normal.
I wrote about this a few years back.

Ok lemme see what I can pull up for notes (I don't have these sorted apparently).

;;;;;;; bios bug update 06/18/14

Set cpu to 42 and ram to 2133
Reset, after post I powered off
Powered back on, set cpu to 43 and ram to 2400
Reset...

No mem bandwith bug.

bios bug 08/28/14
messed around with 125mhz strap, 125mhz bclk
21.33 mem div, 2333mhz, good bandwith
set to 24x mem div, 2666mhz
after saving setting it didn't power cycle, but got bad bandwith
to verify this I messed around with the 1666mhz strap, 166mhz bclk
used 21.33 mem div, 2666mhz
good bandwith
so it's defenitely the 24x mem div

# 08/31/14

original fix + alt fix:
110mhz, assuming 40x cpu

set cpu to 38 (4180mhz)
after save manually power off if it doesn't.
set cpu to 40 (4400mhz) at 110 and set ram div lower at same time
after save manually power off if it doesn't.
set ram div higher and bandwith is good

alt fix:
100mhz, cpu set to 44x, ram set to 2400

set cpu to 42 (4200mhz)
after save manually power off if it doesn't.
set cpu to 44 (4400mhz) and set ram to 2133
after save manually power off if it doesn't.
set ram to 2400, good bandwith

newer fix:
100mhz, mem set to 2400 and cpu 44, or 43 who knows

set mem to 2133
after save manually power off if it doesn't.
set mem to 2400, good bandwith

long and through ver, if it works it might beable to be shortend:

set mem div to 24x
after save manually power off if it doesn't.
set cpu to 35x
after save manually power off if it doesn't.
set cpu to 37x and mem div at 21.33x
after save manually power off if it doesn't.
set mem to 24x
good bandwith?

loaded 114 profile, it power cycled
set mem timings to 10-12-12
it did not power cycle on it's own so I did it manually
after powering back on, set ram div to 24x 2736mhz and mem volts to 1.7v
badnwith was good, bsod'ed on me after the bench but ohwell, it's a ram thing
I have to figure that out later, I haven't tested those speeds yet

....
Blah blah , etc etc.
Just pay attention to how it sets certain settings, if it power cycles a ram div or cpu multi.
And if it didn't power cycle when it was suppose to.
Something like that I think, it's actually been a while.

Only reason I mention it is just incase...

To simplify it, the correct method:

newer fix:
100mhz, mem set to 2400 and cpu 44, or 43 who knows

set mem to 2133
after save manually power off if it doesn't.
set mem to 2400, good bandwith

[Ket]

So you commented on something that wasn't true and then ignored. Ryzen supports 3.0 pcie.

Z270 using the chipset to "create" more pcie lanes doesn't make more lanes. There is still only a 4.0 link to the cpu and 16 actual lanes.

I have never seen that piece of plastic melted on any board ever. I have even had boards that the mosfets have run so hot they have popped. Also remember this is a 95w TDP max board that has the ability to throttle the vrm. I also notice you didn't comment on what the vrm rating was. At all.

Never seen a new GB board that doesn't allow screen capture...

It's still a midrange board hence why it is the M5 and not the M7. I could care less what overpriced amount you spent to buy an item during release month.

I think theres some crossed wires here, so I'll do things bullet style as I'm short on time atm

- You said boards between 200-400 don't have VRM heatsinks with heatpipes, I linked showing they do. My fault for not clarifying thats what the link was for, thought it was obvious enough at the time.
- As I said, I think about absolute worst case scenarios and I just do not like cheap pieces of plastic.
- VRM rating will come when the article is finished, it's stated right at the top it is WIP atm.
- I didn't ignore anything, the comment was for the additional PCI-E 2.0 lanes the X370 chipset provides, not Zen. In other words, why did AMD opt for additional PCI-E 2.0 lanes on the X370 instead of 3.0.
- It was a weird issue, after a DOS re-flash of F5C screen captures started working
- You miss the point, here in the UK 200 dollars translates to around 160 pounds, that is the maximum midrange bracket currency-wise in the UK for this sort of thing. The Gaming 5 isn't worth more than that, hence me bringing up current average prices, you'll also note if you read things that the article says from AWD the board was 170, way below the current average.

Finally, what is with your hostility? No need for it.

For bios caps, if you need to use a cam to doit, set the cam's exposure lv to the lowest it can go, in the neg's.
That will help with taking screens of the monitor.
It increases the shutter speed but also reduces the total light.
I also recommend totally not using flash, because it's showing up in the middle of your pics (monitor pics, the hardware pics are fine).

The plastic shielding thing...
I wish board makers were more serious about this, a meshed metal shield on the board would do fine, and allow heat to escape.

The chipset thing, pcie 3.0 and 2.0.
From what I understand, the cpu is 3.0 16x (actually I think it was 20x), and the chipset is 2.0.
Which is fine for most things, other then maybe m2 cards perhaps (even in that case, "in theory" it should not make a difference).


Your latency bug...
That's an odd one, I don't believe removing the mem should do anything.
However depending on how you boot up settings could make a difference.

Say or ex, you set 2133 mem, and it shuts down to reboot.
After that you set 2666 mem, and say it just resets, it doesn't power cycle, in this case there could be a prob.
Either the mem is gonna be hella fast, stuck with 2133 timings in 2666 mode, or it could go wrong.

My intel has an issue like this, at 100mhz bclk, 2400mhz div is ok, the 2666 div is not ok.
The 2666 has a mem write bandwidth bug.
However, if you manually power cycle it the bandwidth is normal.
I wrote about this a few years back.

Ok lemme see what I can pull up for notes (I don't have these sorted apparently).

*snip*

Don't worry some proper UEFI screens will come screen capture not working was a weird bug, re-flash from DOS sorted it No BCLK adjustment for the Gaming 5 as it lacks a clock gen but thats not a deal breaker. I've also broken out my BIOS tools... first time I've needed to since the P45 days, so I think we know what that probably means

I'll read over your notes fully when I have a bit more time a quick scan of them though looks like we have already tried whats in the notes.

[cdawall]

I think theres some crossed wires here, so I'll do things bullet style as I'm short on time atm

- You said boards between 200-400 don't have VRM heatsinks with heatpipes, I linked showing they do. My fault for not clarifying thats what the link was for, thought it was obvious enough at the time.

Ah see there is the issue. You think those hunk of junk aluminum mono-block heatsinks are improved in any way by a heatpipe...when you say heatpipe cooler this is what I think of

- As I said, I think about absolute worst case scenarios and I just do not like cheap pieces of plastic.

So ALL of the safety features in the board failing, catastrophic failure of mosfets that manages to continue running them even after they have burnt out and what? What is making that plastic shroud melt exactly? I mean if the mosfet has literally caught fire the plastic shroud would quite honestly be the least of my worries.

- VRM rating will come when the article is finished, it's stated right at the top it is WIP atm.

You had the heatsink off its a 30 second google search...

- I didn't ignore anything, the comment was for the additional PCI-E 2.0 lanes the X370 chipset provides, not Zen. In other words, why did AMD opt for additional PCI-E 2.0 lanes on the X370 instead of 3.0.

"It's disappointing to see the X370 only has 8x Gen2 PCI-E, why not Gen3,"

Because you cannot create PCI-e bandwidth. Oddly enough bandwidth wise the 4 lanes supplied by ryzen equals the bandwidth of the 8 2.0 lanes given by the chipset. I would rather like to see what happens when you load all of the lanes provided by z270 to full bandwidth.

- It was a weird issue, after a DOS re-flash of F5C screen captures started working

- You miss the point, here in the UK 200 dollars translates to around 160 pounds, that is the maximum midrange bracket currency-wise in the UK for this sort of thing. The Gaming 5 isn't worth more than that, hence me bringing up current average prices, you'll also note if you read things that the article says from AWD the board was 170, way below the current average.

You have missed what I said. You have purchased a release date product at an inflated release date price. That product is a mid tier board by specs alone. It doesn't even have an external clockgen chip making bus tweaking impossible. The chip is physically gone. In my book regardless of price that moves it down with the worthless products to purchase realm, but feature set as a whole puts it solidly in the mid range.

Finally, what is with your hostility? No need for it.

You were warned of the board quality and allowed your ego to go beyond that point, most of the complaints you have make zero sense and thus the rating you have determined is a bit of a joke.

[chew*]

- Since when does the X370 Gaming 5 UEFI support screen captures? I looked everywhere in F5C and couldn't find any reference for screen captures, I hit the typical hotkeys while in the UEFI with a USB stick in and still nothing then either lol.

I risked killing hardware to show you this by posting a half dead board.

No offense but you need to be more open minded and less stubborn.

fwiw it works on all versions including F1.



FWIW the board is not horrible.

Over simplified. yes.

Bugs. yes. Go read other forums like OCN for all boards....it will make you dizzy

Performance so/so OS changes a lot.

Does it OC. yes.

Does it have half the memory issues of other boards. no

Plug and play tune and go? Not hard to dial in stable settings and be gaming in no time oced. Very Simple.

For someone that just wants to get a quick OC and go this boards for them

For a tuner/tweaker go look elsewhere.

My 2c

EDIT

VRM quality is a non issue, its actually preety decent.

Heatsinks meh. Borrowed from an intel board. Could use improvement

PWM Heat well...... Heatsinks for starters, some may be tuned out up for debate, depends on switching frequency and if they add an option and if currently its tuned to run higher.

[Ket]

Ah see there is the issue. You think those hunk of junk aluminum mono-block heatsinks are improved in any way by a heatpipe...when you say heatpipe cooler this is what I think of

I actually think of heatsinks beefier than that and don't like to see anything that isn't robust, but even on crap heatsinks a heatpipe will help if the components they are cooling are not thermally balanced very well. The heat can at least be spread to the area that is a bit cooler.

So ALL of the safety features in the board failing, catastrophic failure of mosfets that manages to continue running them even after they have burnt out and what? What is making that plastic shroud melt exactly? I mean if the mosfet has literally caught fire the plastic shroud would quite honestly be the least of my worries

It's still a added potential hazard, thats the only point that was being made.

You had the heatsink off its a 30 second google search...

Again, unfinished article. I wanted to put something up sooner rather than later hence the very clear WIP message right at the top of the article. If I had left putting anything up until now it would of had that info along with having all edits already handled by Dan.

"It's disappointing to see the X370 only has 8x Gen2 PCI-E, why not Gen3,"

Because you cannot create PCI-e bandwidth. Oddly enough bandwidth wise the 4 lanes supplied by ryzen equals the bandwidth of the 8 2.0 lanes given by the chipset. I would rather like to see what happens when you load all of the lanes provided by z270 to full bandwidth.

Perhaps I'm not being clear on this one still. I wasn't suggesting creating bandwidth out of thin air I was saying why such a limited number of additional lanes from the X370, seems weird AMD would be ok with the X370 only providing 8 additional lanes.

You have missed what I said. You have purchased a release date product at an inflated release date price. That product is a mid tier board by specs alone. It doesn't even have an external clockgen chip making bus tweaking impossible. The chip is physically gone. In my book regardless of price that moves it down with the worthless products to purchase realm, but feature set as a whole puts it solidly in the mid range.

I didn't buy the board I got it for review, thought I made that clear in the article. We managed to filch one from someone elses supplier Thats why AWD was given credit because we asked who it was. Price is mentioned as a warning not to buy it at current prices. No BCLK adjustment isn't a dealbreaker but it is lame not to have a clock gen chip even if the board settles down to around 150 or so.

You were warned of the board quality and allowed your ego to go beyond that point, most of the complaints you have make zero sense and thus the rating you have determined is a bit of a joke.

I review a board, point out it's good and bad points, and that gives me a ego? All I saw initially from you is hostility.. about a unfinished article. Glad we have moved past that point

I risked killing hardware to show you this by posting a half dead board.

No offense but you need to be more open minded and less stubborn.

fwiw it works on all versions including F1.

I'm always open minded that is why I conduct so many tests and try so many different things to try and isolate the problem, but when people dismiss something without trying to replicate the conditions it's happening under I'm sure you can see how that would be frustrating. Regards the screen captures I said it was either a weird quirk which a re-flash of F5c seems to of sorted or it was due to 4 USB keyboards, all of which work perfectly, not working properly within the UEFI. Basically when the keyboards enter sleep they won't wake up you have to use the mouse to exit the UEFI then re-enter to get the keyboard to wake up and work again.

FWIW the board is not horrible.

Over simplified. yes.

Bugs. yes. Go read other forums like OCN for all boards....it will make you dizzy

Performance so/so OS changes a lot.

Not once have I said the board is bad, just that the UEFI needs a lot of work to bring it up to scratch.

Does it OC. yes.

Does it have half the memory issues of other boards. no

I never said it wouldn't OC, I just don't want to comment too much on it's abilities to OC before at least a sort of final UEFI is available for it, as we all know a new UEFI can make all the difference. Haven't mentioned once it has more memory issues than other boards, just reported the memory latency bug continuously encountered during testing as it's responsible to do so. Could be W10 related as the bug seems to have a mind of it's own with the system sometimes booting and latencies being normal when tested, or might be a compatibility bug. I can provide AIDA screens of the latency bug so you can test it out yourself with W10 but the RipjawsV kit is the only kit we had on hand that would work, rest of the kits definitely aren't compatible.

Plug and play tune and go? Not hard to dial in stable settings and be gaming in no time oced. Very Simple.

For someone that just wants to get a quick OC and go this boards for them

For a tuner/tweaker go look elsewhere.

My 2c

I never said it's not simple to OC, if anything I've said what you have, it's too simple with it's limited options.

EDIT

VRM quality is a non issue, its actually preety decent.

Heatsinks meh. Borrowed from an intel board. Could use improvement

PWM Heat well...... Heatsinks for starters, some may be tuned out up for debate, depends on switching frequency and if they add an option and if currently its tuned to run higher.

Can you clarify on VRM quality? All I can find out is that the VRMs are PowIRstage, can't find out amperage ratings, but if they are anything like NextFET IR3553s they will handle up to 240A. Heatsinks yeah, their not good. I couldn't put it in the actual article I was doing (I went way over the limit as it was ) it would need to be a manual edit for here only but those boron carbide pads along with making the heatsinks make much better contact reduced VRM temps quite a lot.

Anyway gentlemen, I hope that has cleared things up technically related or otherwise, it's easy for things to get misinterpreted

[Ket]

As promised, article has been updated now I've had the chance to test out the latest round of W10 updates so updated accordingly. I also don't have to keep passing this for editing for the small group of boffins I sometimes to things like this for as part of a internal piece when talking alone doesn't cut it so I've been able to word things here and there as I would of if it wasn't for that.

[Ket]

Fixed images, also amended and updated some aspects. Gigabyte have now had 6 months to imprve the UEFI of the board and in that time they have done barely anything as such this revisit to the board and the subsequent amended score is final.