Overclocking and Conclusion
Overclocking
To give a feel for the overclocking performance potential of the X299 AORUS Gaming 3 motherboard, we attempted to push it to known CPU-supported performance parameters with minimal tweaking. At the stock base clock speed of 100Mhz, we pushed the CPU to 4.7GHz with a 3.0GHz mesh bus frequency and 3000MHz memory speeds. This was done at a 1.26V CPU voltage, a 1.20V Mesh voltage, and a 1.355V memory voltage with all other values left at default settings. Unforturnately, the board would not stabilize with memory settings above 3000MHz, even though the memory has run at its full 3200Mhz on other boards. Further, the CPU speed was downclocked to 4.4-4.5GHz with AIDA64 running because of a bug with the AVX code in the BIOS version used for testing. All overclocking sessions remained stable for over 4hrs. System stability was tested running the AIDA64 stability test in conjunction with EVGA's OC Scanner X graphical benchmark running at 1280×1024 resolution and 8x MSAA in stress test mode. Note that 16GB (4 x 4GB) of Corsair Vengeance LPX DDR4-3200 memory modules were used for the overclocking tests
100MHz Base Clock Stats with 5.1GHZ CPU speed
Note that this is is meant only as a quick preview of the board's performance potential. With more time to tweak the settings to a greater extent, pushing to a higher base clock and ring bus speed may have been achievable, in addition to an overnight stability run without issue.
Pricing
As of June 02, the GIGABYTE X299 AORUS Gaming 3 motherboard was available from Newegg.com for $279.99.
Conclusion
The GIGABYTE X299 AORUS Gaming 3 is a solid starting point for the launch of GIGABYTE's Intel X299-based board line. The board features the neutral black board aesthetics, ensuring visual compatibility with most builds in addition to the highly configurable integrated RGB LEDs. The board was well designed with high quality power components used to ensure CPU stability, as well as structural enhancements to the PCIe and memory slots to ensure component stability in a vertical orientation.
The biggest challenge with this board was with overclocking. The AVX issue discussed previously causes the CPU frequency to be downclocked during high utilization scenarios. However, GIGABYTE is currently working on this issue and assures us that an updated BIOS is in the works to address this issue.
We will be taking a more in-depth look into this board and its features in the near future, which should uncover even more strengths with this board and its new branding than was discussed in this preview article.
Strengths
- Stock performance
- Overclocking potential
- Board aesthetics, layout, and design
- Price
- UEFI BIOS design and usability
- Configurable RGB LEDs using RGB Fusion through both UEFI and Windows app
- Dual RGBW headers
- Design of metal reinforced PCIe and memory slots
Weaknesses
- CMOS battery placement
- Lack of third onboard M.2 port
- SATA ports 4 / 5 / 6 / 7 disabled with M.2 drive seated in M.2 port by CPU socket
- Platform maturity
Why is CPU-Z showing 1.92
Why is CPU-Z showing 1.92 vcore @4601MHz – this can’t be accurate. What gives?
CPU-Z is having issues with
CPU-Z is having issues with reading the correct vcore for SKY-X. It is instead reading VccIN which is fed to the FIVR (fully integrated voltage regulator) and will translate into the correct set vcore.
KBL-X doesn’t have FIVR so vcore will be fed directly to the cpu. This is how the switch works between the two vastly different cpus.
HWInfo v554 works well for
HWInfo v554 works well for giving actual voltages and other info. Can be found here:
https://www.hwinfo.com/download.php
I would be leary of this
I would be leary of this board with it’s single 8 pin CPU power as the warning about such boards applies here. My question then though, is that I’ve seen several X299 boards with the 8 pin + 4 pin CPU power and wonder if that’s enough juice to run the VRMs at acceptable temperatures?
They fixed the VRM heat on
They fixed the VRM heat on this model ?
Those CPU seem to require hefty custom liquid cooling, and this eliminate airflow around the socket.
Where the VRM are under heavy stress, and covered with a “heat shield” vs heat sink.
I have a feeling that if you run this board in a closed case overnight running prime95 (small) the PCB will start to degrade around the VRM
All the new X299 mobos have
All the new X299 mobos have the “heat insluator” VRM heatsink issue right now.
Give it a few months and I’d expect it to be sorted.
Fixing the 8 pin connector overheating issues will be harder and take longer I’d expect but is addressable by the end user with a fan or just good case airflow. Still it’d piss me off HAVING to do that stuff on a platform this expensive.
I don’t think it makes much sense at all to buy X299 right now unless you’re rich enough that $300+ is pocket change for you and can easily afford to re-buy parts with fixed issues as they’re released.
Even if you hate AMD and really want X299 now it makes more sense for Threadripper to show up since maybe it’ll end up causing Intel to drops prices some.
As long as you have active
As long as you have active airflow over the VRM sink, you should be ok. That's what I found at least. It doesn't take much airflow either.
Yes I’ve seen other such
Yes I’ve seen other such commentary saying essentially the same thing too but that seems rather half assed to do by default on something as expensive as this platform is. Its the straw breaking this particular camel’s back.
Why be a for-pay beta testing early adopter on a expensive platform now when you can wait a few months and any issues will be fixed for possibly less too?
If Intel had kept the prices down more in general on the chips, done a better IHS, and done some more sane “entry level” X299 CPU’s I’d be forgiving, price and implementation quality matter a whole lot I think, but they half assed all of that so I don’t see why I or anyone should try and meet them half way here.
If it naturally gets that hot
If it naturally gets that hot because of inadequate design, superficially cooling it with fans likely wont be a good long term solution. It would more than likely still fail prematurely.
Not a EE here but all
Not a EE here but all commentary by people who know their stuff suggests the VRM design isn’t bad per se its just the cooling for it was implemented in a piss poor manner that emphasized cosmetics over effectiveness.
Personally I wouldn’t mind going back to the days when mobos only had stuff like colored slots or a non green colored PCB instead of all these LEDs if it meant getting better VRM cooling.
Something like you used to see on the GIGABYTE P35T-DQ6 would be nice to have today now that VRM space is getting cramped due to the large sockets and number of DIMMs on these boards taking up valuable near socket real estate.
Just too much stuff in too small a space putting out too much heat to deal with in the same old crappy manner that they used to be able to get away with on X99 or 270 mobos.
Why are the BIOS settings
Why are the BIOS settings screenshots showing widely off settings? Would’ve been really helpful to show the actual settings used to achieve the overclock – because I’m struggling to get it to work on my Aorus G9 and this might have helped.
For UEFI, I show the max
For UEFI, I show the max settings available in the screencaps.
For the OC, I used the following settings with my 7900X proc:
47 multiplier (CPU)
47 multiplier (Turbo per core)
30 multiplier (Mesh)
100 base clock speed
Mem speed 3000MHz (wouldn't go above that and remain stable)
AVX offset set to 1
VRin – 1.9V
CPU Voltage – 1.26V
Memory Voltage – 1.355V
Mesh Voltage – 1.20V
Enhanced Multicore Performance Enabled
All other settings auto
Hope that helps…
I love you, Morry.
I love you, Morry.
Disappointed they have
Disappointed they have removed the Optical SPDIF output from this $300 motherboard. There is probably a SPDIF header on the motherboard at some location but this will require another card to provide this functionality. There probably arent enough people who use this output to a receiver but its nice to have nonetheless and kind of a cheap move to not include it.
How come he doesn’t mention
How come he doesn’t mention temperatures or power draw? He blames AVX for overclocking problems: “the CPU speed was downclocked to 4.4-4.5GHz with AIDA64 running because of a bug with the AVX code in the BIOS version used for testing.” So why would AVX cause it to downclock? I have read that the AVX circuits are a tremendous heat generator.
We are not getting the full story.
The oc problem was an AVX
The oc problem was an AVX issue with the BIOS. The BIOS would auto downclock the core speed with AVX instructions activer (such as those that run during my AIDA stress testing run). GIGABYTE confirmed this behavior and I have been working with them for a solution. I have not seen this behavior on other X299 boards (like the MSI X299 Gaming M7 ACK that I'm in the process of testing). On the MSI board, all cores run at the same speed while overclocked no matter what stress app I throw at it.
I included HWInfo in the full screen cap of the overclock so that you could see the per core speed vs the core speed that CPUz reports.
Thanks…
279$ no thanks.. ill stick
279$ no thanks.. ill stick with 270 chipset.