Saturday, September 19, 2015

Stream is delayed... and some more

Due to unforeseen issues I'll be streaming on Sunday. Sorry for the delay.

EDIT: More delay on the stream. I start it randomly when I have time some time soon. After the 24th I will go silent as far as blogging goes because I'll be in the UK with no access to any of my stuff. Hopefully I will figure out some way to transport all my stuff and store it.

Friday, September 11, 2015

R9 Fury Tri-X and Fury X Vmod Guide.

Wow am I sloooooow. I promised a Fury Vmod guide in August. It is now September. But hey better late than never. This guide will work for ALL reference PCB Furys. That means any Fury X and the Sapphire Fury Tri-X and the Powercolor Fury. If you have the ASUS STRIX use this guide from Xtreme Addict. Another way to get some extra performance out of the Fury is to try unlock extra cores on it.

VRM overview
The Fury has 4 VRMs. The HBM and Vcore are controlled by the IR 3567B the AUX and 0.95V voltage are controlled by other controllers which I can't find because none of part numbers for the ICs that could be voltage controllers return anything. That also means that I won't be covering how to mod the AUX and 0.95V voltage. If you're on air or water cooling this doesn't matter. For LN2 users you'll have to figure out how to get 1.05V on the 0.95V VRM output your selves because I don't have the resources to figure it out.


DISCLAIMER: Everything past this point will void your warranty! If the GPU stops working due to you doing any of the things described bellow I am at no fault what so ever.

Modding
 Before starting with any volt mods you need a way to measure you voltages. So hook up one wire to any of the GND points, one wire the Core voltage points and one wire to the HBM voltage points. This will allow you to use a multimeter to check the voltages because software does not pick up on the increase in voltage due to volt mods. Also there is no software voltage reading for the HBM or AUX voltages. For AUX it doesn't matter. The stock voltages are 1.2V core and 1.35V HBM.


Now for the actual volt modding. The 3567B is a very very smart voltage controller. It's so smart that if you try to hard mod the power limit you end up with this and run the risk of burning the VRM. Instead just max the power slider in you overclocking software(Sapphire Trixx 5.0.0 goes upto +50% and supports HBM overclocking). To get core voltage under control hook up a 100ohm variable resistor(potentiometer or trimmer) to any of the red points and to ground. To limit the maximum core voltage to 1.8V put a 7ohm resitor in series with the variable resistor. If you want a maximum voltage of 1.5V use a 14ohm resistor. For the HBM hook up a 220ohm variable resistor use a 50ohm resistor to limit the maximum voltage to 1.62V.



Practical tips
Use a fine chisel tip for all the soldering. A needle tip will make everything much harder because the tip of the pen won't actually melt anything.

If you have a temperature regulated iron set it to 240-250C for everything except when trying to solder onto the GND pins of the PCI-e connector. Those act like massive heatsinks so I suggest you raise your pen temperature to 300C.

When you finish soldering one of the connections to an SMD component I recommend using plastidip to hold it in place. Hot glue is harder to apply and very bulky which might lead to you not be able to fit the heatsink back on the card.

Route wires from the 3567B between the PCI-e power connectors. This will hold them in place without the need for more glue/plastidip.

Some results.

Due to budget constraints I do not have 100ohm nor 220ohm potentiometers. So I improvised a dipswitch voltage controller. However they seem to be much higher resistance than the values of the resistors I used to make them. My maximum core voltage is 1.3V and with it I can run 1165mhz core clock this is a 65mhz increase over the 1100mhz that I was getting on stock volts. My maximum HBM voltage is only 1.37V and it didn't improve the HBM overclocking capabilities of my card in the slightest I'm still stuck at 570mhz HBM. So I think HBM doesn't scale with voltage on air cooling or I'm just not giving it enough voltage. Keep in mind that SK Hynix specs HBM to run at 1.2V so over volting it much more than 1.35V might not be the best idea. For the core you can run 1.3V if you keep it sub 80C and if you keep the core sub 60C you can run it at 1.4V. If you set the fans on the Tri-X to 60% you should have no problem keeping the card sub 60C. For benchmarking you can run an extra 100-150mv on top of the safe 24/7 voltage for a given temperature.


Some tips for Tri-X owners
Sapphire used some weird thermal paste on the VRM heatsinks and core. For the core do whatever you usually do when you repaste a GPU but for the VRM heatsinks you can use 0.5mm thermal pads. They will contact the MOSFETs just fine and are cheaper than the 1mm and 1.5mm thermal pads.

Tip to lower or even eliminate coil whine.

Apply hot glue or plastidip on the marked sides of the inductors. If the first time you still get coilwhine you can apply more plastidip/hotglue.

I would like to thank techpowerup.com for allowing anyone to use their PCB pics. My attempts at PCB pictures are better for making desktop backgrounds than Vmod guides:







Thank you to Cooler Master for powering this blog.

Wednesday, September 9, 2015

Sapphire R9 Fury Tri-X OC Physical review.


It's hard to see but Sapphire put a small piece of plastic over the interposer to protect it. It isn't perfect since you can still break the interposer but much better than nothing.



Buildzoid's GPU reviews explained

Specs
3584 stream processors clocked at 1040mhz
4GB of HBM clocked at 500mhz on a 4096bit bus
2 8pin power connectors officially supporting up to 375W
Core clock throttling temperature: 85C°
1 HDMI port3 DisplayPort

Physical Specs
length: 320mm
width: 110mm
height: 2.5 slots

Cooling: 10/10
 When I first played a game on this card I knew I was gonna end up giving away points like they grow on trees. And why wouldn't I the cooling on this card can get the same temperatures as a custom water block. I currently don't have the equipment necessary to do proper noise testing but just ot give you an idea of the fan noise I'll tell you this. With the GPU at 0RPM my sound meter reads about 51.5dB sitting 8cm from the GPU. With the fans at 40% it reads 52 dB. At 55% it gets 59dB. At 75% it reads 66dB and at 100% it reads 73dB. I'd recommend running the card around 55-65% fan speed since that's when the noise temperature trade off is quite reasonable.

VRM: 9/10
Core
- 6 true phases provided by an IR 3567B
- 1 low side IR 6894 MOSFETs providing 105A per phase at 100C case temperature
- 1 high side IR 6811 MOSFET providing 44A per phase at 100C case temperature
- 200KHz up to 2MHz
VRAM
- 1 phase
- 1 low side IR 6894 MOSFETs providing 105A per phase at 100C case temperature
- 1 high side IR 6811 MOSFET providing 44A per phase at 100C case temperature
VAUX
- 1 phase
- 1 low side IR 6894 MOSFETs providing 105A per phase at 100C case temperature
- 1 high side IR 6811 MOSFET providing 44A per phase at 100C case temperature
The low side on this VRM is a beast. However like on the R9 290X the high side here is iffy. If you keep it cool you'll be fine but if you let the VRM get hot you can't let the card pull much more than 500W because you'll be risking burning out the high side. There are plenty of cases of people doing this on the R9 290X. The Fury uses much less power so it's not as much of a risk but keep it in mind if you're going heavy on the voltage and not using LN2 because the VRM can easily overheat. Also do note that in certain situations the card has coil whine. Not enough to be as annoying as say loud fans but enough to be noticeable so keep that in mind. I will try somethings with the card to get rid of the coil whine and will put that into my Fury volt modding post which is coming soon.

Extras:
+1 Dual BIOS
+1 0RPM fans in idle 
+1 VRAM cooling
+1 Backplate
+1 Silicon Interposer protector

Overclocking
The highest I've managed to push this card to was 1144/566mhz using Sapphire Trixx. These were not 100% stable clocks. My 100% stable clocks are 1100/550mhz and require a custom fan curve to keep the GPU sub 55C core temperature. The GPU does not clock well and instabilities are hard to catch. I had the card running perfectly OK at 1135mhz core clock for several hours and then it black screened out of no where when playing a relatively light game. This leads me to believe that the dynamic power management is holding the card back.

Conclusion: 24/20 Why is the score so high?
I love this card. It is by far the best GPU I have ever had. It is way way quieter and cooler than all my past GPUs. It's so good a card that I'm terrified of volt modding it(also because the 3567B has the weirdest voltage sensing circuit I've ever dealt with). Which I'm working on as I finish up this review. I don't really have much to say. I love this card it provides an excellent platform for volt modded overclocking.

The voltmod guide for this GPU is now up

Thank you to CoolerMaster for providing the V1000 PSU that was used for this review.

Tuesday, September 8, 2015

No more Twitch streams

Unfortunately twitch has been a massive pain for my last 2 live stream attempts. So I've switched to Youtube and I'm currently streaming here.

Monday, September 7, 2015

Something Quickly: How to calculate overclocked current and power draw for CPUs and GPUs.

This is what you use to calculate power draw:
(MHzOC/MHzstock)* (VOC2*(TDP/Vstock2))
Just plug in the numbers and you'll get the power draw of the CPU or GPU you're overclocking once it's overclocked. It does have a few issues it only works for specific temperatures and TDP doesn't always give accurate values it's best if you measure the initial power draw rather than relying on the TPD to be accurate. However if you do you power draw and stock voltage measurement when under load and your temperature when overclocked isn't different this will be accurate. A power draw value isn't all that useful when you're trying not to kill a VRM. To get current draw you just use this version of it:
(MHzOC/MHzstock)* (VOC*(TDP/Vstock2))
The power draw value is good for when you're designing a chiller or trying to choose coolers the current draw value is for not killing a VRM. The equation has a +/- 10% margin of error so keep that in mind when using it(AFAIK it does not work with Nvidia Maxwell GPUs at all). It can be much more accurate if I could do more research into it but that requires that I can get my hands on a bunch of CPUs to test it with and several liters of LN2 so I can keep very specific temperatures.

Just for the sake of doing a quick proof: 
 My 3960X measured 353W at 5Ghz using 1.56V running IBT. It has a stock voltage of 1.18V, TDP of 130W and stock clock of 3.3Ghz. Lets plug them in:

(5/3.3)*(1.562(130/1.182))= 344W
So it a little bit of but hey better than nothing.

Friday, September 4, 2015

R9 Fury Unlocking Simplified

As you probably already know it is possible to unlcok the R9 Fury from it's normal 3584 stream processor to 3776 processor, 3840 processor or even 4096 processors at which point it's basically a Fury X.

This guide is a simplification of this guide. When I was doing the process for the first time it took me close to an hour to get through it all. So this is my take on how to make the guide simpler.


First of all download this file  (dropbox mirror) and python 2(Python 3 does not work!). The filecontains all the programs and scripts that you need to unlock your Fury. You need Python to run one of the applications that makes flashing possible.
I hope that none of the software creators have a problem with me putting everything together and re-uploading it. If you do you can just leave a comment and I'll take it down again.

Unzip the folder to where ever you want. Then go to the folder CUinfo and run the program "cuinfo16".
And you'll get to see something like this:
If the last line says something like "X CUs are disabled by HW lock, override is possible at your own risk." You can try flash the card.

In order to flash the card first go into the ATiFlash folder and run the script called "BackupBIOS". This will create a file called: FuryBIOS.rom. Copy and paste this file(DO NOT CUT AND PASTE! THIS IS YOUR WORKING VANILLA BIOS IF THINGS GO SOUR YOU WILL NEED IT!) to the AtomTool folder. Then go back to ATiFlash rename the FuryBIOS.rom to FuryBIOS1.rom and then flip the BIOS switch on the top edge of your Fury. Now run the "BackupBIOS" script again. Once it's done flip your BIOS switch back to it's original position. Now go into the AtomTool folder and run the "makeroms" script. This will creat 3 files called:
3776spBIOS.rom
3840spBIOS.rom
4096spBIOS.rom

Copy these to the ATiFlash folder and run the script called "Flash XXXXSP BIOS" the XXXX stands for how many cores will be unlocked. First run the 3776 core script then restart the computer and check using GPU-z that you now have 3776 cores. If you do stress test the GPU by playing some games or running a stress test. If it's stable you can run the "Flash 3840SP BIOS" script. Then do the same that you did to check that the 3776 BIOS worked. If it does great and you can run the "Flash 4096SP BIOS" script. Then check that it worked the same way you check that the 3776SP and 3840SP BIOS worked.

If at any point the GPU is not stable just go into the ATiFlash folder and run the script that applies the previous BIOS that worked using the corresponding script. If the 4096SP BIOS didn't work flash the 3840SP BIO. If the 3840SP BIOS didn't work flash the 3776SP BIO. If the 3776SP BIOS didn't work run the "Flash Backup BIOS" script.

If the system fails to boot up after a BIOS flash flip the BIOS switch on your R9 Fury and boot up. Once windows is loaded flip the BIOS switch back to the broken BIOS and flash the last working BIOS the same way I explained in the previous paragraph.

If you run into any issues leave a comment down bellow. If you liked this guide please like and share this post potentially sign up for email updates about the blog.

The R9 Fury and Fury X volt modding guide is now up!

Thank you to Cooler Master for powering this blog with their amazing V series power supplies.

Tuesday, September 1, 2015

Gigabyte 78LMT-USB3 Review

Grey and black a color scheme almost as good as black and black. I like it. The board has two 4 pin fan headers one for the CPU and one for a chassis fan. There are six SATA 2.0 ports on the board and 1 IDE connector. Also one internal USB 3.0 port and two USB 2.0 ports.




















four USB2.0, one VGA, one DVI, one HDMI, and two USB3.0 along with basic audio.






















VRM
CPU VCore
- Richtek RT8868 4+1 phase controller
- 4 true phases
- 2 Renesas K0393 Low side MOSFETs rated for 40A at 25C 
- 2 Renesas K03B7 High side MOSFETs rated for 30A at 25C 
- 680nH inductors
- 38.4g Heatsink
- 270KHz to 330Khz 
- 5K rated capacitors however they are very low capacity
- LLC Auto/Regular/Extreme 
- OCP(around 90A) OVP  
CPU NB
- 1 true phase
- 1 Renesas K03B7 High side MOSFETs rated for 30A at 25C 
- 1 Renesas K03B7 Low side MOSFETs rated for 30A at 25C
- 270KHz to 330Khz
- OCP OVP 
RAM
- 1 true phases
- Same MOSFETs as CPU core - No heatsink
- Unknown minimum to unknown maximum KHz
Verdict: 5/10
40A at 25C is very very little. So to be safe it's best to assume a de-rating of 50% at 100C meaning that there is a total of 160A available on the low side. Luckily the RT8868 features OCP so this motherboard shuts down before it catches fire if you push it too far. When under load the capacitors and inductors get into the 110C area without a fan over them. The MOSFETs them selves are kept in the 75-85C area by the heatsink. I'd recommend putting some heatsinks on the inductors and capacitors and a fan over the entire VRM in order to extend life span. Even a basic 50x50mm fan at 2000RPM would help significantly.

BIOS
Voltages:
- Max VCC: 1.4375V
- VCC LLC Extreme(I+75mv/L+150mv) Regular(I+0/L-30mv) 
- Max VNB: 1.475V
- Max VDDR: 1.9V
CPU:
- 35x maximum CPU multi 
- 20x maximum CPUNB multi
- Full active core cuntrol with support 1 core per CU mode
DRAM:
- No XMP Support
- Missing tertiary sub timings
- No support for asymmetrical timings
- All timings are either automatic or manual 
- Up to 2400mhz support 
EXTRAS
- No Extras 
Verdict: 4/10
This BIOS has many flaws. First of all the CPU voltage setting shows incorrect values until you get the OCP to trip. So when you first boot the board and want to overclock the CPU you will see the option to run 1.7125 Vcore however if you actually set that you only get 1.43V when idling and 1.4V while running regular LLC. Extreme LLC is what it says on the box. It's EXTREME LLC. Giving a ridiculous 75mv Vboost at idle and an even more insane 150mv Vboost under load. I suggest you just don't use Extreme LLC because setting up to not trip the OCP the moment you go under load is really really hard and the low idle voltage will mean stability will always be very very iffy when using it. There are also no OC profiles. The postives of the BIOS are the 1 core per CU mode and the relatively easy navigation and the good responsiveness that some of the overly graphical BIOSs these days lack.


That's 214W on the CPU power line. I tried to push more than that but the terrible LLC options and low maximum voltage setting meant that my attempts always pushed well over 240W(was aiming for 220-230W) to the CPU and tripped the OCP. So if you were really brave and patient you cold maybe get a 9000 series FX to run on this board. But you most likely won't and I don't think it's worth the effort.

Testing Results:
The following hardware ran stable, using Intel Burn Test (IBT) for stress testing.
- G,skill ECO 2x2GB 1333 7-7-7-24 1.35V/max clock: 1866mhz 9-10-9-27-2T 1.65V
- Max CPU clock 8 core: 4444mhz at 1.4375V
- Max FSB clock: 217mhz
Verdict: 5/10
My FX 8320E is not a very good chip. And this boards weak VRM will not give you more 
The OCP on the VRM does it's job however it makes rebooting a pain because it doesn't shut the system down it just kills the CPU VRM which leaves the rest of the system still running. When that happens you need to cut AC power from the system because holding the power button doesn't seem to do anything.

Conclusion: 14/30 Points, 0.233pts/$
Between this board and the 970M Pro3 I think the key to choosing the right one is knowing what you want to do. This board is easy to use and very safe. The Pro3 on the other hand offers much more CPU voltage freedom and can definitely push higher clocks but the only safe guard it has is the temperature related throttling. If you just want a basic AM3+ MATX board to get 4.4-4.6Ghz depend on your CPU this board is enough. The VRM has enough cooling to not catch fire and the OCP will stop you before you kill the MOSFETs. You also have an integrated GPU that can save you if you dedicated doesn't want to work. This board also offers better RAM overclocking however that's not really worth talking about with a board that is limited to 1.4375 Vcore. Either way it's a good board and if you want to run an AMD 8 core in an MATX system but don't plan to overclock very much this board has you covered.

If you want to support what I do here please hammer the various share buttons down below. If you want to keep up with the various things I do follow me on Facebook. I will also be putting a video review up on the Youtube channel sometime soon. It is less for being informative and more for comedic value because I think there aren't enough motherboard jokes on the internet.

As always thank you to CoolerMaster for powering the blog.