ASUS X299 SAGE

Just pulled the trigger on one of these. It was an out of the box special. So zero wear and tear at a fraction of the cost of what it was originally worth. I didn't really get it for gaming but I needed something that would do my next personal build justice and my 1070 was getting a little long in the tooth.

​

www.asus.com




Model
KO-RTX3070-O8G-GAMING
Graphic Engine
NVIDIA® GeForce RTX™ 3070
Bus Standard
PCI Express 4.0
OpenGL
OpenGL®4.6
Video Memory
8GB GDDR6
Engine Clock
OC Mode - 1845 MHz (Boost Clock)
Gaming Mode - 1815 MHz (Boost Clock)
CUDA Core
5888
Memory Speed
14 Gbps
Memory Interface
256-bit
Resolution
Digital Max Resolution 7680 x 4320
Interface
Yes x 2 (Native HDMI 2.1)
Yes x 3 (Native DisplayPort 1.4a)
HDCP Support Yes (2.3)
Maximum Display Support
4
NVlink/ Crossfire Support
No
Accessories
1 x Collection card
1 x Speedsetup manual
Software
ASUS GPU Tweak II & GeForce Game Ready Driver & Studio Driver
Dimensions
10.82 x 5.16 x 2.13 in
27.5 x 13.1 x 5.4 Centimeter
Recommended PSU
750W
Power Connectors
2 x 8-pin

Will probably running the studio drivers and wait a while before I start over clocking it. Yes, it will be going in the SAGE X299 build. Yes, I know that's not a PCIe gen 4 board. Even on PCIe 3 this GFX card will out perform an RTX 4060. i don't need bleeding edge. I just need performance.

Pulled the tray on the ATCS 840. All the ball bearings are present. Case is in very good condition. Thanks again, Mr. X. You have restored my faith in overseas shipping. (Part of why I purchased that GFX card. It should be in next month some time.) Also a box came in from NewEgg. It's a 1000 Watt Platinum Certified PSU. (Door crasher special - less than $100) Given the nature of this wonderful case I'll be able to install it before I do anything else. And no, it isn't pink.

Despite the fact that this SAGE build will not be a gaming PC I am not disappointed with my choice in graphics cards.


Further research has revealed that this card will run as well as a 2080 Ti on PCIe 3 and that's good enough for me.

This will be handy for video rendering and content creation. I think I made the right choice.

So what have I done with the SAGE build lately? A: Absolutely nothing.

Well, perhaps not "absolutely" nothing. I got a proper power supply for it to test the GFX card I want to use with it but I have to lug a monitor down the stairs to test it because a regular old monitor won't do it justice. I've ordered some better RAM for it but I want to swap over my good old Platinum Dominator RAM to it because it is tried and tested and proven: A factory matched set of 128 GB of RAM. Pictures forthcoming. In the meantime, here's a rose.

So half the RAM is here and I expect to buy a set each month until I have 128 GB worth. It won't be a full factory matched set but it should work on the X99 just fine so that I can move the factory matched set of Samsung B die RAM over to the SAGE. I have the RTX 3070 on the SAGE now but I have to install a different PSU on the test bench. We decided that 1000 Watts will look pretty in pink, but it isn't going in the Cooler Master and it won't be going in the Level 10 GT.

I should have entitled this thread the SAGE SAGA.


Well, it's a good thing I had no intentions of keeping the Vetroo PSU as a permanent fixture as I got it all plugged in on the test bench and nothing...

I mean, nothing. Zip. Nada. Checked all my connections many times over. It don't work.

Tried it on a different system board because it didn't have the 6 pin connector for the CPU I needed. Still nothing.

Tried it with a jumper: Dead to the world.

Admittedly, it does look pretty in pink once it was installed.

In fact, I would say it looked 'Instanisnlretty'.

But i was stopped dead in my tracks when I found out the CPU connector was for either 4 or 8 Pins and the SAGE needed a 6 pin connector.



Now I do realize there are work arounds for this but if I can't even get the thing to power up a regular ATX board it isn't likely I'll be able to get it to power up an EATX either. :::SIGH::: Back to the drawing board.

So I went over the connectors again, tried the jumper on the ATX board, and viola! Got the Pink Vetroo working again. Alas, on the wrong system board. Then again, it wasn't going to be a permanent thing. Just wanted to put some xtra power on the test bench for the GFX card. Meanwhile I found this so perhaps I can get away with not using the 6 pin connector at all.

Well, domestics seem to be keeping this old man away from his favourite hobby but I managed to squeeze in a little time to work on the SAGE work station board. Admittedly, I don't mind losing one PCIe slot to a thick GFX card but losing three is simply unacceptable. The whole idea behind having 44 lanes is to be able to use them so I decided to eliminate the cause of the issue: The shroud.



Since I really don't need all the RGB glitz as it is going inside a Thermal Take Level 10 GT case with a leetle tiny top window on the side panel I decided to de-shroud the RTX 3070 GFX card. It seems to work rather well. The huge 240 mm fan on the side panel of the case should provide ample air cooling.


I noticed that I can even access the PCIeX16_2 slot with a short little riser cable for just in case more PCIe connectivity is needed. This would mean that I will have access to all 7 PCIe slots.

The preferred position for my Thunderbolt 3 card was blocked by the shroud and this is also where I plug in the cable for TB3. The RGB shroud simply had to go. The ASUS RTX 3070 KO card is more toward the higher end of the spectrum for NVIDIA 3070 graphics cards. I plan on using it in conjunction with my Thunderbolt 3 card which also plugs into the display port on the graphics card. Alas, this shroud covers the cable port for the Thunderbolt card making it pretty much inaccessible. I don't have this issue on my X99 board as the Thunderbolt header is not located under a PCIe slot on that particular system board. Also with my NVIDIA 1070 graphics card I only lose one PCIe slot (instead of two) as it is not nearly so thick.

So... I thought about whining about this in the ASUS section but since it didn't affect all ASUS builds I'll keep it here for now.

I just lost all my drivers for my 24H2 after running and installing Windows Update and then I hit a wall on a security update. So, after numerous attempts at getting it to take I opted for the stand alone version which led to the April 4 24H2 repair installation. Lots of fun. Suddenly Windows decided that I no longer had TPM2 on the X299 Windows 11 ready system board. Somewhere along the lines of updates the TPM option was disabled in the UEFI and so the trouble shooter decided my PC was not Windows 11 compatible.

The Fix:

The fix for Windows thinking your PC isn't TPM2 ready is to default the entire BIOS, turn off the PC, and jumper the CMOS. After doing this (and of course placing back the jumper) you have to go back into your BIOS and look for the TPM settings (They're under the Thunderbolt settings — make sure you scroll all the way down or you may not see them). Just to make life interesting TPM settings on the SAGE X299 are called PTT. I remember this because on the initial Win 11 installation I had to hunt for them to activate TPM2 and wasted more time trying to locate this setting than I care to admit. The page may look as though it is completely scrolled down to the bottom and that the last line of text is for your Thundebolt settings but no, there is one more line that is hidden from view if the window isn't scrolled down all the way.

Once you have done this you can reboot your PC and go back to your OS where you can enjoy the privilege of convincing Windows that your PC really does have TPM2. This will require a Windows 24H2 stand alone repair that you can get from Windows Catalogue. *Warning* The repair takes time. It takes more time than reinstalling Windows 11 so forget about being in a rush. After this you can proceed to go back into your UEFI and make all those adjustments you lost because you had to "reset your BIOS". This of course will be followed by having to re-install drivers that Windows has magically forgotten existed once upon a time. I'm not there yet. I'll save re-installing the drivers for another time. But for those who are struggling with the X299 SAGE WORK STATION because Windows doesn't recognize TPM2. . .

Okay, so after hours of attempting to train the brand new Crucial RAM on the X299 WS SAGE I gave up. It seems the SAGE just doesn't like it. Half the time only two of the sticks will show and sometimes I can get half of it to show. Oftentimes the system will give ye olde RAM warning *Beep-beep* and I gotta tell ya this isn't exactly uber high profile RAM. This is good old solid Micron at its finest and I can't figure out what the deal is. If I could get another set of the TEAM GROUP XTREEM T-FORCE RAM that I used in my AMD build I would because the board seemed to just love that. Alas, they don't make that RAM anymore (which is a shame because it is extremely easy to overclock). So I pulled all 128GB of the stuff and popped it into my X99 and it booted up without a hiccough. No problems, 128GB show up just fine and it's faster than the Dominator RAM that was originally in the old girl. So, it looks like the SAGE is getting the B die Dominator RAM instead, which is kind of sad because although it is good RAM it's also slower. Not that this should really affect much.



Now bear in mind that the above is not the SAGE. All the RAM appears to be in good working order although I haven't run MemTest86 on it yet. That's coming. Yes, yes the CPU is showing up in WinInfo as a nice, toasty red but it's solid as a rock and it runs stable. It is too bad about HVCI but if I enable it then I'm going to have issues with TBMT all over again and it will likely mess up my dual boot so I'm leaving it alone. Other than that my non-compliant Windows 11 abomination seems to be chugging along just fine. Methinks the X99 board is a much nobler beast than the X299.

So although the adapter card for the NVMe does seem to function it doesn't seem to do well on the test bench. I had to remove the screw to the bracket to keep a full connection. Later on I moved it to the the native socket as it seemed to do better there. I'm guessing this issue will cease once I get the board in the case. The short little riser cable is connected to slot #2 and it works. No substantial loss in functionality with a cable this short. I just haven't decided what to use it for yet. Slot #3 is reserved for Thunderbolt 3 and the cable attachment for Thunderbolt is just above it. A better cooler is on order.



I'm hoping to use slot 4 (the one populated with a TEAMGROUP stick of NVMe in the picture) and slot 5 for RAID hardware. Slot 7 is reserved for USB2.0 because the case has connectivity for this and the system board has none. Sometimes USB2.0 is all you really need. The idea of a work station is keeping options open.



The X299 absolutely abhors the Micron RAM I purchased for it. I fought with it for hours, training it in the usual A-1, C-1 method suggested in the manual and it was a no go. I pulled the low latency Corsair "Platinum" Dominator RAM from the X99 and it trained wonderfully in the X299 without a hitch. I might increase the RAM voltage later on, after I get a decent cooler. The difference between this Samsung B die RAM and the Crucial RAM is astounding. Micron does make good RAM but it isn't very forgiving on this platform. The Crucial Pro RAM had to go to the X99 board instead.

looking good.
best of luck, Steve ..

Thanks, Steve. I'm on my AMD build as I'm currently running scans on the memory to eliminate which stick of RAM is giving me trouble. It's a sad day in the PC world when Micron starts issuing defective RAM. I suppose it only goes to show that it never pays to be a 'fan boi' of any name brand. This is supposed to be high quality XMP RAM. Perhaps I got a bad batch.

Sorry for the blurry picture. My hands aren't as steady as they used to be and I don't believe I can do screen shots from BIOS. I'm 'shooting in the dark' as one person here put it, but I'm zeroing in on finding specifically which stick or sticks are bad. Micron generally makes excellent RAM and this saddens me to see. At least now I know why my system shut down and reported an overclocking failure after years of running a stable OC. This might also explain why I had so much trouble training this brand new RAM to show in all 8 DIMM slots. Only 33 of 48 tests passed. Hopefully I'll be able to return it.

So far, so good. Two more passes to go. Full test takes around 20 hours. IF all checks then I will know which set of RAM is bad. A simple process of elimination.

Test completed. I now have the defective RAM isolated and I've established that the remaining RAM is in good condition.

I'll be returning the defective RAM and ordering another two sticks of the same. Hopefully it will be in better shape.

So this is the stuff that failed on me. Half of it worked and half of it didn't. So far I've sent back 64 GB worth and that really surprises me because Micron generally makes excellent memory. (Micron makes Crucial.) Just recently I recall stating that I've never had to send back a stick of Micron memory in all my PC building years and lo, and behold, it has happened! Not just once, but twice. Amazon will be getting a review.




Admittedly, this is entry level XMP RAM despite the "Pro Series" logo, rated at 3200MHz. Fast enough for this old man considering I plan on using 128GB of the stuff. I also purchased a new cooler for the X299.

Amazon.ca

Rave reviews and although it isn't quite an NH-D15 it comes about two or three degrees close in thermals. Good enough for the girls I go out with. It's a multi-award winner and at just a little over $50 Canadian (tax included) I'd say the price is right. This means that the X99 will keep the Noctua cooler and that will be going into the Cooler Master ATCS 840.

The hold up with the BEAST is that I can't put the cooler on without the RAM in first. Well, I suppose I could but it would be a pain. This is not a teeny-tiny cooler. This is huge. Think NH-D15 huge. So I'm beginning to think I'll just have to roll with the Dominator Platinum RAM (which isn't as fast) and ye olde RAM fans after all. This will block off some of the intake on the new cooler but it can't be helped. My memory needs to stay cool. The RAM does look nice but it'll be hidden under some ugly cooling fans.







At least in the case of the Corsair RAM all DIMMs show. This was not the case with the Micron RAM. Hopefully I'll be able to ramp up the Corsair memory to at least 2666 MHz as I did on the X99. The saving grace with this RAM is that it has excellent latency and I had those numbers down 14-14-14-30 without increasing the voltage when it was in the X99. Hopefully I'll be able to get those numbers again on the X299 once I get the Peerless Assassin cooler installed. Where on earth do they come up with such ridiculous names??

So I'm running the RAM scan on the X99 with all DIMM slots populated. 128GB of Micron memory that is supposedly identical. Not sure why the last DIMM shows as "MICRON" while the rest all show as "Crucial". This old X99 can't seem to handle the full 3200MHz clock speed so I had to dial it down a bit. Yes, I suppose I could up the voltages but if it passes the MemTest 86 scan I want to transfer it to the X299 SAGE and snag back the Dominator RAM for the X99. So far, so good. One more pass after this one.

This is the sort of thing one likes to wake up to first thing in the morning:



Once upon a time, when I was a younger, wealthier, man I had the paid version of this software. I have no idea where it went but the free version works fine for my needs and purposes. I was under the impression that it only ran 11 tests but apparently it runs 14 now and I'm fine with that. So 14 tests, 4 passes.


I apologize for the green background. This is an anomaly from my KVM switch, which I'm too lazy to fix. It doesn't affect the information though. Once I have the system board pulled, cleaned, and set up correctly, these temperatures should drop. (Although they aren't exactly drastic @52C) Still, I like to keep my RAM cool and this does not meet my personal standards for adequate cooling. I may have to look into some individual coolers for memory as my Corsair cooler fans will will not be going into the ATCS 840 build. In fact, as things progress, I'm really beginning to wonder if I shouldn't put the SAGE X299 into the ATCS 840 after all. It could save me a little time.


I guess I'm slipping in my old age because I'm pretty sure MemTest 86 ran a total of 12 (0-thru 11) tests before. Perhaps they upgraded. That's fine by me.



The freeware version is fine for determining if your RAM is any good. I'm running all these tests on my X99 platform but I hope to be transferring the RAM to the SAGE platform in due time. I'm a little torn betwixt and between which system board to put in the ATCS 840 as the old X99 platform actually has native USB2 headers for the ports on that case. The SAGE does not. This is not really a big deal as it can be remedied with an expansion card. I suppose I'm a bit of a 'purist' in this regard as I greatly value my PCIe lanes. I know, in the long run, utility will win me over.


This concludes my walk thru with MemTest 86 (Free Version). Above is more of the features you get with the paid version. (Still need to find that key. lol) Honestly, old age is killing me I'm sure. Anyway, the RAM works. Hopefully I'll be able to get those latencies down but I'm not expecting any miracles as it is not Samsung B die. Not that Micron memory is anything to sneeze at. I'm just saying...