*Official Benchmark Thread* - Post it here or it didn't happen :D

[Papusan]

Yup. Only the 4090 are able to keep up it's value. We will see what happens closer to Christmas holidays.

 

 

 

------------------------------------------------------------------------------------------------

 

intel XMPs and motherboard QVLs are a useless mess. The QVL list will never be correct as long the testers (elite benchers) at the different brands continue use binned hardware in their testing and validating of firmware. Once again why use binned HW? So the brand can brag about unreliable results that are not rooted in reality. Only to sell overpriced HW. See from 10:15

 

[D2ultima]

12 hours ago, jaybee83 said:

 

nah, 6000 was never the max ryzen 7000 could support, it was just always touted as the "sweet spot" in terms of fire and forget easy tuning and bang per buck. same for Intel at 7200, basically. you can go up to roughly 6400, in rare cases to 6600 in gear 1. remember, intel cant even use gear 1 at all with ddr5, its always gear 2 going from 4800 all the way to 8000+. switching to gear 2 on ryzen 7000 also allows to go 8000+. as with intel, not every mobo or cpu imc can take it, but in general its definitely possible.

Huh... I heard lots from various outlets that you basically can't go past 6000MHz on AMD without fclk tweaking and a ton of trial and error, and it's also pretty board/CPU dependent. Didn't sound like a sweet spot to me, more of like a limit unless you want to dedicate days to it

[electrosoft]

6 hours ago, D2ultima said:

Huh... I heard lots from various outlets that you basically can't go past 6000MHz on AMD without fclk tweaking and a ton of trial and error, and it's also pretty board/CPU dependent. Didn't sound like a sweet spot to me, more of like a limit unless you want to dedicate days to it

 

Before a major update enabling what is effectively "Gear 3" for AMD, AMD memory clocking was limited to ~6000-6600 and a proper ratio is tied to FCLK (1:3) and MEMCLK (1:2). This is basically Gear 2 for AMD and was new territory for Intel with 11th gen offering 1st gen Gear 2 albeit on memory that couldn't handle high enough frequencies to make it make sense (aka DDR4) so you stuck with G1 like their previous CPUs. These are carryovers from AM4 with a bit of enhancements for AMD. Now we can effectively open up DDR5 to 1:4 and 1:3 and hit the same memory frequencies as Intel.

 

 

 

Let me pre-empt this by stating outside of posting big fancy bandwidth benchmarks, I really haven't seen anything major gains wise in what is effectively G3 (4? 🤔)  (since by nature it runs G2 aka 1:2 MCLK:MEM natively. AMD doesn't run 1:1) for AMD with the ability to scale memory up into Intel range. I posited this many months ago when the new memory mode was announced but untested and that it may take next gen AMD 8000 series CPUs or later to really take advantage of it like Intel does with 12th-14th gen and even then Intel leaves a nice chunk of real world performance on the table as you scale past 7200 but still yields tangible gains especially compared to AMD in real world (IE not synthetics) testing. Real world, both scale pretty badly but this will improve with time. AMD just does it worse atm. AMD right now is in Intel 11th gen G2 mode in regards to their G3 and taking advantage of high speed DDR5. It is there and it works but can't really utilize it properly yet outside of throwing up bandwidth numbers showing the memory is indeed running at and processing the expected number of MT/s.

 

6000 is the "sweet spot" in regards to the natural ratio of AM5 (MCLK:MEM 2:1, FCLK:MEM 3:1) and all IMCs able to hit it 99.99% of the time depending on memory configuration. Almost all AM5 chips can't keep a proper ratio past ~6400 (2133 fclk) on mem and many top out at 6200 (2066 fclk). You can run it async if you want out of the 1:3 ratio (mclk 1:2). Running memory below 6000 is gimping your AM5 platform and at the very minimum this is what you shoot for and tighten your timing as much as possible. My fclk tops out at 2066 but I found testing 6000 up to 6200 sync and 6600 async that 6000 super tight sync provides the best performance that is actually utilized.

 

Still, I'm glad it's there because eventually we will see meaningful performance gains if not this generation then potentially next as it is here to stay....same as Intel where hopefully Arrow Lake can better utilize all that bandwidth DDR5 has to offer than Alder or Raptor.

 

I still need to pick up some competent 8000+ sticks to play with between both my AMD platform and MSI Z790i Edge 2-dimmer + 13900KS. I just haven't done the ground work or asked @Mr. Fox and @tps3443 what are their current flavor of sticks. 🙂

 

 

[Raiderman]

1 hour ago, electrosoft said:

 

Before a major update enabling what is effectively "Gear 3" for AMD, AMD memory clocking was limited to ~6000-6600 and a proper ratio is tied to FCLK (1:3) and MEMCLK (1:2). This is basically Gear 2 for AMD and was new territory for Intel with 11th gen offering 1st gen Gear 2 albeit on memory that couldn't handle high enough frequencies to make it make sense (aka DDR4) so you stuck with G1 like their previous CPUs. These are carryovers from AM4 with a bit of enhancements for AMD. Now we can effectively open up DDR5 to 1:4 and 1:3 and hit the same memory frequencies as Intel.

 

 

 

Let me pre-empt this by stating outside of posting big fancy bandwidth benchmarks, I really haven't seen anything major gains wise in what is effectively G3 (4? 🤔)  (since by nature it runs G2 aka 1:2 MCLK:MEM natively. AMD doesn't run 1:1) for AMD with the ability to scale memory up into Intel range. I posited this many months ago when the new memory mode was announced but untested and that it may take next gen AMD 8000 series CPUs or later to really take advantage of it like Intel does with 12th-14th gen and even then Intel leaves a nice chunk of real world performance on the table as you scale past 7200 but still yields tangible gains especially compared to AMD in real world (IE not synthetics) testing. Real world, both scale pretty badly but this will improve with time. AMD just does it worse atm. AMD right now is in Intel 11th gen G2 mode in regards to their G3 and taking advantage of high speed DDR5. It is there and it works but can't really utilize it properly yet outside of throwing up bandwidth numbers showing the memory is indeed running at and processing the expected number of MT/s.

 

6000 is the "sweet spot" in regards to the natural ratio of AM5 (MCLK:MEM 2:1, FCLK:MEM 3:1) and all IMCs able to hit it 99.99% of the time depending on memory configuration. Almost all AM5 chips can't keep a proper ratio past ~6400 (2133 fclk) on mem and many top out at 6200 (2066 fclk). You can run it async if you want out of the 1:3 ratio (mclk 1:2). Running memory below 6000 is gimping your AM5 platform and at the very minimum this is what you shoot for and tighten your timing as much as possible. My fclk tops out at 2066 but I found testing 6000 up to 6200 sync and 6600 async that 6000 super tight sync provides the best performance that is actually utilized.

 

Still, I'm glad it's there because eventually we will see meaningful performance gains if not this generation then potentially next as it is here to stay....same as Intel where hopefully Arrow Lake can better utilize all that bandwidth DDR5 has to offer than Alder or Raptor.

 

I still need to pick up some competent 8000+ sticks to play with between both my AMD platform and MSI Z790i Edge 2-dimmer + 13900KS. I just haven't done the ground work or asked @Mr. Fox and @tps3443 what are their current flavor of sticks. 🙂

 

 

I have a set for you, check your PM@electrosoftJust collecting dust on my desk. I would rather you have them, and share your settings. Its an inexpensive set to tinker with.

 

[Papusan]

On 8/4/2023 at 12:07 AM, Papusan said:

I know you could make it better 🙂

 

 

@electrosoft Will you still try out your CM adapter replacement or have you asked for the v1.1 ? And how smart.... A re-call would destroy profits and reputation @Rage Set Better send out an email with coupon code so you can get vr.1.1 for free. Cheaper as well.

 

What's the difference between stop sell it for now vs re-call? @ryan Do your math and you can clearly see whats going on in Papusan's head🙂

 

 

 

 

 

Nice. Cablemod adapter v1.1 is out and now it seems they prepare for a 1.1-v2.0, LOOL

https://youtu.be/2UPdeFEy0zA?si=prmXa6sK1WH1bvMw&t=172

 

The new adapter v1.1 is thighter than the v1.0 burning mess but still not thight enough and will pull out 9:05

 

[tps3443]

2 hours ago, electrosoft said:

 

Before a major update enabling what is effectively "Gear 3" for AMD, AMD memory clocking was limited to ~6000-6600 and a proper ratio is tied to FCLK (1:3) and MEMCLK (1:2). This is basically Gear 2 for AMD and was new territory for Intel with 11th gen offering 1st gen Gear 2 albeit on memory that couldn't handle high enough frequencies to make it make sense (aka DDR4) so you stuck with G1 like their previous CPUs. These are carryovers from AM4 with a bit of enhancements for AMD. Now we can effectively open up DDR5 to 1:4 and 1:3 and hit the same memory frequencies as Intel.

 

 

 

Let me pre-empt this by stating outside of posting big fancy bandwidth benchmarks, I really haven't seen anything major gains wise in what is effectively G3 (4? 🤔)  (since by nature it runs G2 aka 1:2 MCLK:MEM natively. AMD doesn't run 1:1) for AMD with the ability to scale memory up into Intel range. I posited this many months ago when the new memory mode was announced but untested and that it may take next gen AMD 8000 series CPUs or later to really take advantage of it like Intel does with 12th-14th gen and even then Intel leaves a nice chunk of real world performance on the table as you scale past 7200 but still yields tangible gains especially compared to AMD in real world (IE not synthetics) testing. Real world, both scale pretty badly but this will improve with time. AMD just does it worse atm. AMD right now is in Intel 11th gen G2 mode in regards to their G3 and taking advantage of high speed DDR5. It is there and it works but can't really utilize it properly yet outside of throwing up bandwidth numbers showing the memory is indeed running at and processing the expected number of MT/s.

 

6000 is the "sweet spot" in regards to the natural ratio of AM5 (MCLK:MEM 2:1, FCLK:MEM 3:1) and all IMCs able to hit it 99.99% of the time depending on memory configuration. Almost all AM5 chips can't keep a proper ratio past ~6400 (2133 fclk) on mem and many top out at 6200 (2066 fclk). You can run it async if you want out of the 1:3 ratio (mclk 1:2). Running memory below 6000 is gimping your AM5 platform and at the very minimum this is what you shoot for and tighten your timing as much as possible. My fclk tops out at 2066 but I found testing 6000 up to 6200 sync and 6600 async that 6000 super tight sync provides the best performance that is actually utilized.

 

Still, I'm glad it's there because eventually we will see meaningful performance gains if not this generation then potentially next as it is here to stay....same as Intel where hopefully Arrow Lake can better utilize all that bandwidth DDR5 has to offer than Alder or Raptor.

 

I still need to pick up some competent 8000+ sticks to play with between both my AMD platform and MSI Z790i Edge 2-dimmer + 13900KS. I just haven't done the ground work or asked @Mr. Fox and @tps3443 what are their current flavor of sticks. 🙂

 

 

I think the best value is the G.Skill 7200’s 2x24GB Hynix 3GBit m-die kit. I was super frustrated with my current ram just a few days ago, however, I updated my bios and lowered my expectations with timings because it is 48GB after all, and I’ve hit gold with my current sticks. Previously, I’d run some crazy high gaming stable overclock of like 8800Mhz+ on my ram it would be game stable but not actually stable stable, but now I’ve started pushing for “Actually stable settings”  which is new to me. But I finally have that magical DDR5 8600 truly stable on my 2x24GB 7200 sticks with only 1.500V VDD/ 1.450V VDDQ. These are only $200 dollar sticks which ain’t bad for a 48GB kit. I was also running another DDR5 8000C34 profile that was very fast and also actually stable. I cannot get 8800+ stable for more than about 8 minutes on TM5. But then again, I’m also not all that great with memory overclocking and timings etc. 

 

You can kind of go both ways with this kit. High speed higher timings or lower speed with some much much tighter timings.

 

As far as I can tell, I do not think the newer faster memory kits offer any better bang for buck. Just a ton of money wasted lol. 
 

If you can find the 8000 kit for around the same price or not much more, I’d also go for those too! I think it’s all about the same stuff though and luck of the draw. 
 

[Raiderman]

32 minutes ago, tps3443 said:

I think the best value is the G.Skill 7200’s 2x24GB Hynix 3GBit m-die kit. I was super frustrated with my current ram just a few days ago, however, I updated my bios and lowered my expectations with timings because it is 48GB after all, and I’ve hit gold with my current sticks. Previously, I’d run some crazy high gaming stable overclock of like 8800Mhz+ on my ram it would be game stable but not actually stable stable, but now I’ve started pushing for “Actually stable settings”  which is new to me. But I finally have that magical DDR5 8600 truly stable on my 2x24GB 7200 sticks with only 1.500V VDD/ 1.450V VDDQ. These are only $200 dollar sticks which ain’t bad for a 48GB kit. I was also running another DDR5 8000C34 profile that was very fast and also actually stable. I cannot get 8800+ stable for more than about 8 minutes on TM5. But then again, I’m also not all that great with memory overclocking and timings etc. 

 

You can kind of go both ways with this kit. High speed higher timings or lower speed with some much much tighter timings.

 

As far as I can tell, I do not think the newer faster memory kits offer any better bang for buck. Just a ton of money wasted lol. 
 

If you can find the 8000 kit for around the same price or not much more, I’d also go for those too! I think it’s all about the same stuff though and luck of the draw. 
 

Bro @electrosoftjust got bang for the buck 😁

[electrosoft]

39 minutes ago, Raiderman said:

Bro @electrosoftjust got bang for the buck 😁

 

Thanks Bro for the deal!

 

Phase 1 is going to be seeing where they top out in classic G2 AMD mode to see if the timings can get any tighter than the A-die 6400 sticks I have in there now at 6000/6200.

 

Then G3 for 8000 AMD

 

Then  MSI Z790i Edge + 13900ks testing since I know this combo can hit 8200 tested with my buddy's Geil sticks a few months back no problem. Of course this combo has been sitting in the MSI box for the last month or so since the Hyte Revolt 3 was a bust and I haven't had the time to spec, prep and build out in the Cougar Dust 2 which is still sitting in the box unopened.

 

 

 

 

[jaybee83]

10 hours ago, D2ultima said:

Huh... I heard lots from various outlets that you basically can't go past 6000MHz on AMD without fclk tweaking and a ton of trial and error, and it's also pretty board/CPU dependent. Didn't sound like a sweet spot to me, more of like a limit unless you want to dedicate days to it

 

welp...its RAM tuning, of course its finicky 😄 no news there haha. what makes it a bit more complicated is if ure used to Intel RAM tuning and switch to AMD after a long time on team blue (like i did), ull have a bit of a learning curve since things work a bit differently. but in the end its all the same: u can shoot for a quick n dirty oc and call it a day or tinker away for weeks n months to get every single timing tightened to the max with lowest possible voltages. and the higher u go, the more tinkering and fine tuning it needs, exactly the same for intel. can do a quick n dirty 7200 or spend tons of time to tune in 8000+.

 

on that subject: i just nailed my 6200 settings on gear 1, ill post some data on it later. now ill need to transfer my bios profiles to the new version after i update. then off i go check and see if i can get 8000 running finally! was a goood decision to dial in 6200 first, now i have a better idea of what timings 8000 will likely need to be stable 🙂 

 

3 hours ago, electrosoft said:

 

Before a major update enabling what is effectively "Gear 3" for AMD, AMD memory clocking was limited to ~6000-6600 and a proper ratio is tied to FCLK (1:3) and MEMCLK (1:2). This is basically Gear 2 for AMD and was new territory for Intel with 11th gen offering 1st gen Gear 2 albeit on memory that couldn't handle high enough frequencies to make it make sense (aka DDR4) so you stuck with G1 like their previous CPUs. These are carryovers from AM4 with a bit of enhancements for AMD. Now we can effectively open up DDR5 to 1:4 and 1:3 and hit the same memory frequencies as Intel.

 

 

 

Let me pre-empt this by stating outside of posting big fancy bandwidth benchmarks, I really haven't seen anything major gains wise in what is effectively G3 (4? 🤔)  (since by nature it runs G2 aka 1:2 MCLK:MEM natively. AMD doesn't run 1:1) for AMD with the ability to scale memory up into Intel range. I posited this many months ago when the new memory mode was announced but untested and that it may take next gen AMD 8000 series CPUs or later to really take advantage of it like Intel does with 12th-14th gen and even then Intel leaves a nice chunk of real world performance on the table as you scale past 7200 but still yields tangible gains especially compared to AMD in real world (IE not synthetics) testing. Real world, both scale pretty badly but this will improve with time. AMD just does it worse atm. AMD right now is in Intel 11th gen G2 mode in regards to their G3 and taking advantage of high speed DDR5. It is there and it works but can't really utilize it properly yet outside of throwing up bandwidth numbers showing the memory is indeed running at and processing the expected number of MT/s.

 

6000 is the "sweet spot" in regards to the natural ratio of AM5 (MCLK:MEM 2:1, FCLK:MEM 3:1) and all IMCs able to hit it 99.99% of the time depending on memory configuration. Almost all AM5 chips can't keep a proper ratio past ~6400 (2133 fclk) on mem and many top out at 6200 (2066 fclk). You can run it async if you want out of the 1:3 ratio (mclk 1:2). Running memory below 6000 is gimping your AM5 platform and at the very minimum this is what you shoot for and tighten your timing as much as possible. My fclk tops out at 2066 but I found testing 6000 up to 6200 sync and 6600 async that 6000 super tight sync provides the best performance that is actually utilized.

 

Still, I'm glad it's there because eventually we will see meaningful performance gains if not this generation then potentially next as it is here to stay....same as Intel where hopefully Arrow Lake can better utilize all that bandwidth DDR5 has to offer than Alder or Raptor.

 

I still need to pick up some competent 8000+ sticks to play with between both my AMD platform and MSI Z790i Edge 2-dimmer + 13900KS. I just haven't done the ground work or asked @Mr. Fox and @tps3443 what are their current flavor of sticks. 🙂

 

 

 

nice sum up bud, kudos 🙂 

[tps3443]

25 minutes ago, electrosoft said:

 

Thanks Bro for the deal!

 

Phase 1 is going to be seeing where they top out in classic G2 AMD mode to see if the timings can get any tighter than the A-die 6400 sticks I have in there now at 6000/6200.

 

Then G3 for 8000 AMD

 

Then  MSI Z790i Edge + 13900ks testing since I know this combo can hit 8200 tested with my buddy's Geil sticks a few months back no problem. Of course this combo has been sitting in the MSI box for the last month or so since the Hyte Revolt 3 was a bust and I haven't had the time to spec, prep and build out in the Cougar Dust 2 which is still sitting in the box unopened.

 

 

 

 

How come you have not been running that binned ks more? Is it slower than the Ryzen chips in the games you play? 

[electrosoft]

4 minutes ago, tps3443 said:

How come you have not been running that binned ks more? Is it slower than the Ryzen chips in the games you play? 

 

Here is why the SP115 has been sitting in a box the last few months LOL:

 

Remember my idea was to build out a banger sff system which is why I scooped up that 13900KS to have a killer 13900 level portable system if possible. I picked up a Hyte Revolt 3 w/ PSU as my "welcome to SFF building" experience.

 

I jumped through all these hoops to get it working in the Hyte Revolt 3, but the Hyte mounts the rad directly over the motherboard and the pressure of the tubing would eventually cause the mount to destabilize and temps would eventually creep up from expansion and contraction.

 

I should have known when I first tried to use an extra on my shelf EVGA CLC 280 which has tubing right out of the top = even more downward pressure when closed. In a normal test case to make sure the hardware was sound, temps were absolutely fine for days on end (IE, high 70s for a 41.5-42k CB23 run with this SP115 NO delid on a 280mm AIO w/ KPx lol....insane). As soon as I mounted it and closed it temps immediately spiked to 95+. Rinse/Repeat = same results.

 

I switched to a pump side tubing mount AIO I had in the closet (ID Cooling) mid level to low level 280 AIO. Temps were a touch higher (~82 max) because it isn't as good as the EVGA CLC 280 (which isn't even the best for a 280) but it would always initially run fine but over the course of hours or days, temps would eventually spike back up to 95+. I'd take it out, remount, same cycle. Finally, I remounted and left the side open causing no pressure on the pump/mb and it ran fine for days on end. I took everything out and tested it on the table, ran fine no problems.

 

So the issue was the Hyte design along with the downward pressure on the MSI Z790i. As a test, I put my daughter's 12400 in there and re-installed it into the Hyte, closed it up and yep a day or two later, temps were spiking into the high 80s / low 90s.

 

I dismantled everything and put it all back into their boxes and then I've been preoccupied with 4 laptops I have (or had) here: NH55 (which I had been meaning to get back to for some time to tame a desktop 12900k in it but the SFF phase one took over), Second Asus G18 (testing, benchmarking, trying several sets of memory, then returned because of AC BS), my daughter's P377SM-A which she finally wanted to downsize from for college so I switched her to an Acer Swift 16 OLED AMD laptop I had on the shelf (which is amazing) so I was tinkering with her system and some spare parts (4940MX, couple of MXM GPUs) for benchmarking.

 

Now I have a lull in the action, so I'm turning my attention back to the whole SFF thing as I have a Cougar Dust 2 sitting here ready and waiting which takes a traditional approach to AIO mounting up top. I also will probably give a Dan A4 a whirl too. In the interim, I did some research on 240 AIOs and picked up an EK 240mm which seems to work the best in the Cougar and A4 (which also does a traditional on top AIO) in regards to tube management along with having some of the best thermals out there and a low profile block. It has been sitting sealed in its box for over a month now.

 

I will rebuild everything in my test case first with the 8000 sticks from @Raiderman along with the EK 240mm and re-qualify everything then build out in the Cougar Dust 2 (aka SFF adventures Vol 2).

 

I like the Hyte Revolt 3, but if or when I get back around to building out in it, it will definitely be an air cooled system for the CPU so something like a 7800X3D would rock in there. That leaves a lot of PSU room open with the included 700w PSU for something like a 7900xtx for a killer, portable AMD advantage system.

 

 

[tps3443]

30 minutes ago, electrosoft said:

 

Here is why the SP115 has been sitting in a box the last few months LOL:

 

Remember my idea was to build out a banger sff system which is why I scooped up that 13900KS to have a killer 13900 level portable system if possible. I picked up a Hyte Revolt 3 w/ PSU as my "welcome to SFF building" experience.

 

I jumped through all these hoops to get it working in the Hyte Revolt 3, but the Hyte mounts the rad directly over the motherboard and the pressure of the tubing would eventually cause the mount to destabilize and temps would eventually creep up from expansion and contraction.

 

I should have known when I first tried to use an extra on my shelf EVGA CLC 280 which has tubing right out of the top = even more downward pressure when closed. In a normal test case to make sure the hardware was sound, temps were absolutely fine for days on end (IE, high 70s for a 41.5-42k CB23 run with this SP115 NO delid on a 280mm AIO w/ KPx lol....insane). As soon as I mounted it and closed it temps immediately spiked to 95+. Rinse/Repeat = same results.

 

I switched to a pump side tubing mount AIO I had in the closet (ID Cooling) mid level to low level 280 AIO. Temps were a touch higher (~82 max) because it isn't as good as the EVGA CLC 280 (which isn't even the best for a 280) but it would always initially run fine but over the course of hours or days, temps would eventually spike back up to 95+. I'd take it out, remount, same cycle. Finally, I remounted and left the side open causing no pressure on the pump/mb and it ran fine for days on end. I took everything out and tested it on the table, ran fine no problems.

 

So the issue was the Hyte design along with the downward pressure on the MSI Z790i. As a test, I put my daughter's 12400 in there and re-installed it into the Hyte, closed it up and yep a day or two later, temps were spiking into the high 80s / low 90s.

 

I dismantled everything and put it all back into their boxes and then I've been preoccupied with 4 laptops I have (or had) here: NH55 (which I had been meaning to get back to for some time to tame a desktop 12900k in it but the SFF phase one took over), Second Asus G18 (testing, benchmarking, trying several sets of memory, then returned because of AC BS), my daughter's P377SM-A which she finally wanted to downsize from for college so I switched her to an Acer Swift 16 OLED AMD laptop I had on the shelf (which is amazing) so I was tinkering with her system and some spare parts (4940MX, couple of MXM GPUs) for benchmarking.

 

Now I have a lull in the action, so I'm turning my attention back to the whole SFF thing as I have a Cougar Dust 2 sitting here ready and waiting which takes a traditional approach to AIO mounting up top. I also will probably give a Dan A4 a whirl too. In the interim, I did some research on 240 AIOs and picked up an EK 240mm which seems to work the best in the Cougar and A4 (which also does a traditional on top AIO) in regards to tube management along with having some of the best thermals out there and a low profile block. It has been sitting sealed in its box for over a month now.

 

I will rebuild everything in my test case first with the 8000 sticks from @Raiderman along with the EK 240mm and re-qualify everything then build out in the Cougar Dust 2 (aka SFF adventures Vol 2).

 

I like the Hyte Revolt 3, but if or when I get back around to building out in it, it will definitely be an air cooled system for the CPU so something like a 7800X3D would rock in there. That leaves a lot of PSU room open with the included 700w PSU for something like a 7900xtx for a killer, portable AMD advantage system.

 

 

Why not just go with a custom loop with the revolt case? Would that work? 

[electrosoft]

2 hours ago, tps3443 said:

Why not just go with a custom loop with the revolt case? Would that work? 

 

If I was going with a custom loop (most likely 280+240) it would be in something like an NR200 or Dan C4. For a single rad custom loop, it would be an A4.

 

Hyte would run into the same problems and it really, for all its bulk, just isn't optimized for a good custom loop. There's a reason there aren't many custom loop build outs for it 🙂

 

Take a look at this one.

 

I wouldn't have even bothered. I would have switched to a NR200 MAX, N2, C4 or even an A4 for a custom loop. In hindsight, for an air cooled build I would go with a Fractal, Sliger or S400. A sandwich build for an air cooled SFF seems perfect.

 

But it's a fun learning process. The Hyte taught me many things I don't want one of them being, for an AIO model, the AIO must mount in a traditional manner up top and I also don't like recessed access to the I/O ports on the MB.

 

The Cougar Dust 2 seems to check all the boxes for a semi portable SFF build but we will see. The one I picked up I got at half the cost of new and it comes with Noctua fans instead of stock POS.

 

 

 

 

 

 

[Mr. Fox]

Congrats @electrosoftthat is a sweet deal on the memory kit Brother @Raidermanblessed you with.

 

So, I spent the ENTIRE weekend (literally) rebuilding my loop and integrating the 5 gallon tank and chiller. Got the tank re-insulated. I totally dismantled the MO-RA 360 and cleaned the fans (4 years of grime build-up... was a chore) meticulously with Clorox wipes and Q-tips, added two new 10-port fan hubs on each side. I was going to delid the 13900KS I got from Germany, but I am too spent, LOL. As of now, about 20 hours in total and 30 feet of clear tubing later. Even with the chilled water circulating through the radiator, my idle temps are between 9° and 15°C now as long as the fans are turned off. It took a LONG time to get 5 gallons of water chilled versus 15 to 20 minutes with the volume of the loop without the 5 gallon tank.

 

I added a 200mm fan to the back of the chiller as a booster to its internal condenser fan and it made a notable difference in the amount of hot air getting pushed out of it. I could barely feel anything coming out with the internal fan and now hot air just pours out of this sucker. Now I need to snip the wires to those crazy ugly blue LEDs. When I redid the cooling lines this time I gave myself an extra 5 feet of length so I can roll the chiller out from under the desk (it is on a dolly) when I want to vacuum or sweep the floor.

[D2ultima]

1 hour ago, Mr. Fox said:

Congrats @electrosoftthat is a sweet deal on the memory kit Brother @Raidermanblessed you with.

 

So, I spent the ENTIRE weekend (literally) rebuilding my loop and integrating the 5 gallon tank and chiller. Got the tank re-insulated. I totally dismantled the MO-RA 360 and cleaned the fans (4 years of grime build-up... was a chore) meticulously with Clorox wipes and Q-tips, added two new 10-port fan hubs on each side. I was going to delid the 13900KS I got from Germany, but I am too spent, LOL. As of now, about 20 hours in total and 30 feet of clear tubing later. Even with the chilled water circulating through the radiator, my idle temps are between 9° and 15°C now as long as the fans are turned off. It took a LONG time to get 5 gallons of water chilled versus 15 to 20 minutes with the volume of the loop without the 5 gallon tank.

 

I guarantee you, I would never have the patience to keep my PC down for as long as it'd take to build something like that, but at the same time, MAN would I love to use such a chill PC

[Mr. Fox]

1 minute ago, D2ultima said:

I guarantee you, I would never have the patience to keep my PC down for as long as it'd take to build something like that, but at the same time, MAN would I love to use such a chill PC

That is my play PC. I also have my work computer that is wicked enough in its own rights. And, my Dell Precision turdbook. So, I was not without a PC, but I was so engrossed in what I was doing I didn't have time to use the other systems.

 

Below is my work computer (Banshee in signature) in the opposite corner of this tiny room. Excuse the mess. I still haven't 100% finished cleaning up from the project.

 

 

[cylix]

Now this is interesting.  NVIDIA entering PC Chip market. Going ARM way, also AMD is going to release ARM architecture chips

 

https://www.reuters.com/technology/nvidia-make-arm-based-pc-chips-major-new-challenge-intel-2023-10-23/

[Rage Set]

33 minutes ago, cylix said:

Now this is interesting.  NVIDIA entering PC Chip market. Going ARM way, also AMD is going to release ARM architecture chips

 

https://www.reuters.com/technology/nvidia-make-arm-based-pc-chips-major-new-challenge-intel-2023-10-23/

 

There has been rumblings of this since a little before Nvidia tried to buy ARM. Nvidia was trying to get away from buying CPUs from AMD and Intel for their Enterprise servers. 

 

The question is, can Nvidia do what Qualcomm couldn't (well their next Nuvia based CPU is supposed to rival Apple's M series of processors)? Ultimately, all of these companies - AMD, Intel, Qualcomm and perhaps Nvidia, are at the mercy of Microsoft. No matter how great the chips are, if MS doesn't produce a version of Windows that can utilize the power of the chips, then it is a non-starter. I am waiting for one of these companies to offer their own Linux based OS to compliment their chips...then MS and in turn their competitors are in trouble.  

[D2ultima]

15 hours ago, Mr. Fox said:

That is my play PC. I also have my work computer that is wicked enough in its own rights. And, my Dell Precision turdbook. So, I was not without a PC, but I was so engrossed in what I was doing I didn't have time to use the other systems.

 

Below is my work computer (Banshee in signature) in the opposite corner of this tiny room. Excuse the mess. I still haven't 100% finished cleaning up from the project.

 

 

Ahhh if I had a second device I would definitely consider it. Well I do have my old P870 but I gave it to my brother in law to use since he doesn't have a gaming PC, and he's enjoying it.

[Papusan]

More picture for the pict lover @electrosoft😀 I couldn't resist due it's price. I bought an 1060 card un-seen and I didn't even ask for the brand model when I clicked the buy button. And guess what I got.... Asus 1060 Strix O6GB. And with Samsung vram. Damn stupids at Asus @Mr. Fox They gave their cheaper 1060 OC cards Samsung vram but put in the less wanted Micron vram for their more expensive 1070 Strix OC............

 

 

I paid $30 USD for it. Not bad. I put it into the test machine. Run a bit too hot so it need re-paste. The card is never re-pasted because I see the woid warranty sticker is still on top of one of the screws.

 

 

[tps3443]

43 minutes ago, Papusan said:

More picture for the pict lover @electrosoft😀 I couldn't resist due it's price. I bought an 1060 card un-seen and I didn't even ask for the brand model when I clicked the buy button. And guess what I got.... Asus 1060 Strix O6GB. And with Samsung vram. Damn stupids at Asus @Mr. Fox They gave their cheaper 1060 OC cards Samsung vram but put in the less wanted Micron vram for their more expensive 1070 Strix OC............

 

 

I paid $30 USD for it. Not bad. I put it into the test machine. Run a bit too hot so it need re-paste. The card is never re-pasted because I see the woid warranty sticker is still on top of one of the screws.

 

 

The GTX1070’s are just such an amazing value is all, AIB’s we’re using anything in place of GDDR5, to meet that demand lol. They were beating the GTX980Ti by 3% in all games at any resolution and were selling like hot cakes left and right. Even peoples grandmas were buying them up simply because they were such a “Good value” 😂 

 

Anyways, funny how Micron is the best now? Or at least my 3090 Kingpin has Micron, and boy do them things overclock like no bodies business. And even with our DDR5, Samsung sucks in todays standards! 🤣
 

It’s so weird, because you wanted Samsung ram with a GTX1060/GTX1070 any GTX1060/1070 with Micron was considered a loser.. But the best GPU memory was the new GDDR5X stamped on the new GTX1080 which was SURPRISE SURPRISE Micron and 10,000Mbps, and it overclocked pretty darn good! Then back around with the 2080Ti’s Samsung is best lol. Which I had a 2080Ti FE with Samsung and it was a GREAT clocker.
 

Tech makes no sense at all to me.. 🙂

 

 

[electrosoft]

6 minutes ago, tps3443 said:

The GTX1070’s are just such an amazing value is all, AIB’s we’re using anything in place of GDDR5, to meet that demand lol. They were beating the GTX980Ti by 3% in all games at any resolution and were selling like hot cakes left and right. Even peoples grandmas were buying them up simply because they were such a “Good value” 😂 

 

Anyways, funny how Micron is the best now? Or at least my 3090 Kingpin has Micron, and boy do them things overclock like no bodies business. And even with our DDR5, Samsung sucks in todays standards! 🤣
 

It’s so weird, because you wanted Samsung ram with a 1060/1070 and Micron sucks. But the best GDDR5 was the new GDDR5X stamped on the new GTX1080 which was also Micron and 10,000Mbps. 
 

Tech makes no sense at all.. 🙂

 

 

 

Just like DDR4 and Samsung B-die was king but then DDR5 everyone wanted Hynix.....

 

 

[Papusan]

22 minutes ago, tps3443 said:

The GTX1070’s are just such an amazing value is all, AIB’s we’re using anything in place of GDDR5, to meet that demand lol. They were beating the GTX980Ti by 3% in all games at any resolution and were selling like hot cakes left and right. Even peoples grandmas were buying them up simply because they were such a “Good value” 😂 

 

I mean 1060 also sold in loads. They have been on top in Steam a long time. Just sad see Asus put up the middle finger for their best 1070 OC card and gave some of them Micron. Remember Asus put out many different SKUs of same tier. At least give their overpriced Strix OC cards the best vram. But nope. It's Asus @Mr. Fox🤔

 

22 minutes ago, tps3443 said:

Anyways, funny how Micron is the best now? 

 

Because only Micron can offer GDDR6X. Nvidia partnered with Micron for their high end SKUs in 3000 and 4000 series. 

 

Edit. https://videocardz.com/newz/micron-24-gbps-gddr6x-memory-for-geforce-rtx-40-series-is-now-in-production

[tps3443]

1 hour ago, Papusan said:

 

I mean 1060 also sold in loads. They have been on top in Steam a long time. Just sad see Asus put up the middle finger for their best 1070 OC card and gave some of them Micron. Remember Asus put out many different SKUs of same tier. At least give their overpriced Strix OC cards the best vram. But nope. It's Asus @Mr. Fox🤔

 

 

Edit. https://videocardz.com/newz/micron-24-gbps-gddr6x-memory-for-geforce-rtx-40-series-is-now-in-production

It’s wouldn’t say it’s “Asus”. It’s probably has to do with meeting demand is all. I remember the first batches of Pascal had Samsung, but the 2nd wave of 1070’s to come in-stock had Micron on some cards. I think the micron met the spec and it was used as an alternative GDDR5 no?

[Papusan]

30 minutes ago, tps3443 said:

It’s wouldn’t say it’s “Asus”. It’s probably has to do with meeting demand is all. I remember the first batches of Pascal had Samsung, but the 2nd wave of 1070’s to come in-stock had Micron on some cards. I think the micron met the spec and it was used as an alternative GDDR5 no?

Asus as well as other brands shouldn’t use the worst chips for their flagship products. That’s totally wrong. You pay massive premium for their best products and get chips from the bottom barrel isn’t great. Rather wait throw it out if there is shortage of the better chips/ICs. That’s on Asus and the rest of the greedy companies. They need to stop with this.