PC Building Guide and Discussion #14

[Osprey]

Pretty excited to see some in-depth reviews for the 7800XT myself (after work). At that price and with 16GB it will be much more tempting than AMD's flagship even if less future proof.

They were only just announced and I don't think that any reviewers have them yet, so we may have to wait a week or two for reviews.

I agree with your thoughts, though. I've been wanting a 7900 XT or XTX for a few months, but I've really hesitated to spend $750-950 on a GPU. $500 is much easier to justify, especially since I can sell my current card for about half that, meaning that the upgrade would cost me only $250. If I were to buy the 7900 XT at $750, I'd be paying $500, or twice as much, to upgrade to a card that's probably only 10-20% faster. The 7800 XT might not be that big of improvement over what I have (at least not with ray tracing on), but I would honestly pay a couple hundred just for 16GB, since having only 8 is starting to worry me. I'll wait for the reviews before I decide, but the 7800 XT is tempting at that price.

 

[aleshemsky83]

I don’t want to be a buzzkill about FSR3, but you all know it’s gonna suck right?

 

[Mikeaveli]

I don’t want to be a buzzkill about FSR3, but you all know it’s gonna suck right?

AMD reveals long-awaited FSR 3 tech and frame gen for every DX11/DX12 game

DF goes 'eyes on' with FSR 3 - and a new AMD frame-gen solution that works with every DX11 and DX12 game.

www.eurogamer.net

According to DF its looking pretty good, image quality is similar to DLSS 3 like I predicted. If they can get input lag down to similar levels as DLSS 3 thats a huge win but I have a feeling that will only be the case for titles that officially support FSR3. The input lag on titles using the driver implementation (Fluid Motion Frames) will probably be much worse, though if they somehow make it reasonable that would be incredible for all PC gamers.

 

[aleshemsky83]

https://www.eurogamer.net/digitalfoundry-2023-amd-reveals-long-awaited-fsr-3-tech-and-frame-gen-for-every-dx11dx12-gam

According to DF its looking pretty good, image quality is similar to DLSS 3 like I predicted. If they can get input lag down to similar levels as DLSS 3 thats a huge win but I have a feeling that will only be the case for titles that officially support FSR3. The input lag on titles using the driver implementation (Fluid Motion Frames) will probably be much worse, though if they somehow make it reasonable that would be incredible for all PC gamers.

That’s what they say from their initial first look, and tbh DF is usually overly optimistic. FSR 1.0 was essentially a sharpening filter that wasn’t even better than turning the sharpness up on your monitor. FSR 2.0 while not completely useless isn’t better than a lot of games regular TAA. DF gave both of those good reviews. I tend to lean towards a hardware unboxed for a more blunt look at these things. DF kind of gets caught up in the shiny new toys.

 

[SolidSnakeUS]

Question for you guys on what you'd do or what you'd recommend.


I am about to upgrade my system (been sitting on parts for a week to find the time to do it). I already have a 7700X (can't be returned) and I got the ASRock B650E Taichi Lite, RAM, etc. I also bought 6 140mm Noctua NF-A14 Chromax fans to replace the EK Vardar RGB fans in my case and for the EK AIO Elite 280mm that I have. I also have a spare 240mm EK Nucleus Dark that has yet to be used (opened the box but never used). My 280mm AIO is over 2 years, going on 3 now (Nov ish will be 3), while the 240mm was gotten only a few months ago. Do you think I should just use the newer 240mm for the system or do you think I should re-use the 280mm AIO and just use up all 6 140mm fans that I grabbed? Thanks in advance!

 

[#37]

Question for you guys on what you'd do or what you'd recommend.


I am about to upgrade my system (been sitting on parts for a week to find the time to do it). I already have a 7700X (can't be returned) and I got the ASRock B650E Taichi Lite, RAM, etc. I also bought 6 140mm Noctua NF-A14 Chromax fans to replace the EK Vardar RGB fans in my case and for the EK AIO Elite 280mm that I have. I also have a spare 240mm EK Nucleus Dark that has yet to be used (opened the box but never used). My 280mm AIO is over 2 years, going on 3 now (Nov ish will be 3), while the 240mm was gotten only a few months ago. Do you think I should just use the newer 240mm for the system or do you think I should re-use the 280mm AIO and just use up all 6 140mm fans that I grabbed? Thanks in advance!

This is gathered from my experience of dealing with systems that ran too hot (AMD). Your mileage may vary and you might have to experiment a bit. Or, depending on the case, none of this but the theory will apply...

**If you are running a gaming rig where everything is being overclocked and you need 4 fans to keep it from shutting off, you would be better served going to liquid cooled... Likely the voltage usage of half a dozen fans is going to affect your overall performance (FPS). The goal is efficiency. Heat and voltage loss are the enemies of performance. Having too many fans working inefficiently is not an ideal solution.

But if you aren't... here are the basics of designing an air cooled system for a home built computer.

A push-pull system of 2 case fans is ideal. You want one pulling air into (intake) the front of the machine and one pushing air out the back (exhaust) so that you get a constant and steady flow of air through the case. Assuming this is a tower, the fans should be placed in opposite corners if possible, intake low and exhaust placed high. The key is having a constant and steady air flow.

Also, I have found that it's better to let the bios manage fan speed, as running fans wide open by default can actually cause the machine to run hotter due to a starvation of air pressure entering at the front panel, as this area is usually a bottleneck that can only allow so much air volume to enter the system, with the rear fan 'trying' to expel air faster than the front fan can pull it in... actually decreasing meaningful air flow across components. And that would be my concern with trying to use 6 fans, the loss of efficient air flow (But I suppose it depends on the case). The idea is to create a steady volume of air flow that moves through the case efficiently. Creating too much air volume is bad, and can congest the air flow. Creating too little air volume is bad, as the system becomes starved of a constant and steady flow.

Really, not knowing your intended use for the machine or anything about your case, I would start out with one fan pushing air out the back and see how it goes. If it is running hot, add an intake fan at the front. Use the best thermal compound available to you, let price not be an object. As best you can, run the wiring in a way that wont cause obstructions in air flow to heat emitting components or the exhaust fan. Card placement plays a role too. If it still runs hot. get a better cpu cooler. If it still runs hot, get a better case as some cases breath better than others. Avoid any case that has a plastic like coating on the outside (it acts as an insulator), as they tend to retain heat where a plain old metal case lets heat dissipate quicker. A fancy looking high-end gaming case with a plexi-glass door (insulator) and interior light looks pretty cool, but was probably designed to be cooled with liquid, not air.

Oh, and make sure the machine can breath. Depending on case design, sitting the case directly onto a carpeted floor might obstruct air intake if the front vent is on the bottom of the front panel.

Also, invest in a PSU that has way more watts than you will ever need (especially if you go with 6 fans..). The idea is that it is never under any real stress and shouldn't contribute to the creation of excess heat. It 'should' also last a very long time and not fail, and, thus, take your motherboard with it.

The main problem with AMD chips is that they run at a higher voltage and therefore create more heat than an Intel, with the added bonus of less performance (Ok, depending on use). (Disclosure: This is one reason I will never buy another AMD chip.)

Kudos to anyone who bothered to read all that, I didn't actually intend to write a paper.

 

[Osprey]

The main problem with AMD chips is that they run at a higher voltage and therefore create more heat than an Intel, with the added bonus of less performance (Ok, depending on use). (Disclosure: This is one reason I will never buy another AMD chip.)

That hasn't been the case since AMD introduced the first Ryzen processors 6 years ago. They've been very comparable and often even better than the Intel offerings.

Here's an LTT video from 4 years ago showing AMD running cooler and with better performance than the comparable Intel chip:


Here's a HUB video from late last year showing the same thing with two of the top chips from the current generations:


I set both videos to the relevant timestamps.

In competing with AMD's Ryzen processors, Intel has been massively increasing their chips' power consumption. That has led to greater heat, and though it's also led to higher clocks than AMD, they can't be sustained for very long and the Intel chips end up thermal throttling, which actually results in less performance (seen in the renderings in both videos taking longer on Intel, despite higher clock speeds).

 

[#37]

That hasn't been the case since AMD introduced the first Ryzen processors 6 years ago. They've been very comparable and often even better than the Intel offerings.

Here's an LTT video from 4 years ago showing AMD running cooler and with better performance than the comparable Intel chip:


Here's a HUB video from late last year showing the same thing with two of the top chips from the current generations:


I set both videos to the relevant timestamps.

In competing with AMD's Ryzen processors, Intel has been massively increasing their chips' power consumption. That has led to greater heat, and though it's also led to higher clocks than AMD, they can't be sustained for very long and the Intel chips end up thermal throttling, which actually results in less performance (seen in the renderings in both videos taking longer on Intel, despite higher clock speeds).

Though interesting and informative, I am not convinced by those videos. The problem, for me, is that they are rather linear and only testing one aspect. If you are a gamer, the target audience, that is what you are wanting to see. Fair enough. (We both know I could come back with a presentation where Intel comes out looking better, but that's not really what I am interested in.) I'd like to see tests that are running a digital audio workstation with 50-100+ tracks, reverbs, and midi synths. Or a linux system compiling huge packages like LVM or even GCC from source, like how a package maintainer might use it.

Hmm, here is an interesting link showing a completely different kind of test. Spoiler, the Xeon won.. but that isn't the point, the point is it is a test created from a non-gaming perspective.

AMD Ryzen 3900X vs Intel Xeon 2175W Python numpy - MKL vs OpenBLAS

In this post I've done more testing with Ryzen 3900X looking at the effect of BLAS libraries on a simple but computationally demanding problem with Python numpy. The results may surprise you! I start with a little bit of history of Intel vs AMD performance to give you what may be a new...

www.pugetsystems.com

Anyway, I am willing to concede that, much like other two horse races, the situation is fluid from generation to generation. No technological lead has proved insurmountable, which is quite good for the consumer.

I can only speak from my own experiences, granted it isn't the latest generation of stuff (and not qualifying my statement with that info was a mistake) and I didn't mention the actual application of the hardware when criticizing performance (another oopsie).

 

[PeteWorrell]

My last two builds have been Intel but it's simply a fact that recent generations of AMD chips have been more power efficient in different applications even outside of gaming.

AMD use chiplets that run on more modern designs thanks to TSMC. Intel rely on their own internal foundries that are behind on a technological level. They are trying to catch up but in the meantime they pile on more cores and voltages that makes their CPUs very powerful but not very efficient.

 

[Osprey]

Though interesting and informative, I am not convinced by those videos. The problem, for me, is that they are rather linear and only testing one aspect. If you are a gamer, the target audience, that is what you are wanting to see. Fair enough. (We both know I could come back with a presentation where Intel comes out looking better, but that's not really what I am interested in.) I'd like to see tests that are running a digital audio workstation with 50-100+ tracks, reverbs, and midi synths. Or a linux system compiling huge packages like LVM or even GCC from source, like how a package maintainer might use it.

Both of the videos, at the timestamps that I linked to, showed the AMD chips beating the Intel chips in performance and thermals in multi-core rendering workloads in Blender and Cinebench, which are the kinds of benchmarks that you want to see if you're concerned with productivity. You might find similar tests that show Intel outperforming AMD, but what's been pretty consistent for the last 5 years or so is that Intel is the one that runs hotter under extreme stress, often at 100C. The point is, if Intel runs a bit hotter and the two chip makers trade blows when it comes to performance, then it's really not true anymore that AMD is hotter and slower than Intel.

Hmm, here is an interesting link showing a completely different kind of test. Spoiler, the Xeon won.. but that isn't the point, the point is it is a test created from a non-gaming perspective.
https://www.pugetsystems.com/labs/h...on-2175W-Python-numpy---MKL-vs-OpenBLAS-1560/

That's an outrageously unfair comparison, though. A 14-core workstation Xeon processor with AVX512 extensions, 128GB of RAM and a workstation GPU is naturally going to outperform a 12-core consumer Ryzen processor with no AVX512 extensions, only 16GB of RAM and a consumer GPU.

 

[PeteWorrell]

Both of the videos showed the AMD chips beating the Intel chips in multi-core rendering workloads in Blender and Cinebench, which are the benchmarks that you want to see if you're concerned with production work.

That's an outrageously unfair comparison. Naturally, a 14-core workstation Xeon processor with AVX512 extensions, 128GB of RAM and a workstation GPU are going to beat a 12-core consumer Ryzen processor with no AVX512 extensions, only 16GB of RAM and a consumer GPU.

I didn't click on that link but wow that's terrible testing. I don't know how you could take those results seriously unless you are seriously biased against AMD.

 

[SolidSnakeUS]

This is gathered from my experience of dealing with systems that ran too hot (AMD). Your mileage may vary and you might have to experiment a bit. Or, depending on the case, none of this but the theory will apply...

**If you are running a gaming rig where everything is being overclocked and you need 4 fans to keep it from shutting off, you would be better served going to liquid cooled... Likely the voltage usage of half a dozen fans is going to affect your overall performance (FPS). The goal is efficiency. Heat and voltage loss are the enemies of performance. Having too many fans working inefficiently is not an ideal solution.

But if you aren't... here are the basics of designing an air cooled system for a home built computer.

A push-pull system of 2 case fans is ideal. You want one pulling air into (intake) the front of the machine and one pushing air out the back (exhaust) so that you get a constant and steady flow of air through the case. Assuming this is a tower, the fans should be placed in opposite corners if possible, intake low and exhaust placed high. The key is having a constant and steady air flow.

Also, I have found that it's better to let the bios manage fan speed, as running fans wide open by default can actually cause the machine to run hotter due to a starvation of air pressure entering at the front panel, as this area is usually a bottleneck that can only allow so much air volume to enter the system, with the rear fan 'trying' to expel air faster than the front fan can pull it in... actually decreasing meaningful air flow across components. And that would be my concern with trying to use 6 fans, the loss of efficient air flow (But I suppose it depends on the case). The idea is to create a steady volume of air flow that moves through the case efficiently. Creating too much air volume is bad, and can congest the air flow. Creating too little air volume is bad, as the system becomes starved of a constant and steady flow.

Really, not knowing your intended use for the machine or anything about your case, I would start out with one fan pushing air out the back and see how it goes. If it is running hot, add an intake fan at the front. Use the best thermal compound available to you, let price not be an object. As best you can, run the wiring in a way that wont cause obstructions in air flow to heat emitting components or the exhaust fan. Card placement plays a role too. If it still runs hot. get a better cpu cooler. If it still runs hot, get a better case as some cases breath better than others. Avoid any case that has a plastic like coating on the outside (it acts as an insulator), as they tend to retain heat where a plain old metal case lets heat dissipate quicker. A fancy looking high-end gaming case with a plexi-glass door (insulator) and interior light looks pretty cool, but was probably designed to be cooled with liquid, not air.

Oh, and make sure the machine can breath. Depending on case design, sitting the case directly onto a carpeted floor might obstruct air intake if the front vent is on the bottom of the front panel.

Also, invest in a PSU that has way more watts than you will ever need (especially if you go with 6 fans..). The idea is that it is never under any real stress and shouldn't contribute to the creation of excess heat. It 'should' also last a very long time and not fail, and, thus, take your motherboard with it.

The main problem with AMD chips is that they run at a higher voltage and therefore create more heat than an Intel, with the added bonus of less performance (Ok, depending on use). (Disclosure: This is one reason I will never buy another AMD chip.)

Kudos to anyone who bothered to read all that, I didn't actually intend to write a paper.

It's more an upgrade for my machine that I'm keeping parts from.

Kept my Gigabyte RTX 3080, 2TB NVMe, 4TB NVMe, 4TB HDD, case and Corsair RM850x 80+ Gold. Everything is going swimmingly with using the 240mm EK Nucleus Dark. Lowest temps I've gotten were 37C and highest were 91C so far.

I did fully replace 4 of the case fans with 4 of the Noctua NF-A14 Chromax fans and I'm using the 2 120mm fans that came with the 240mm and this thing has been f***ing whisper quiet. To the point where I keep thinking my PC is actually turned off.

Seriously, jumping from the 3700X to the 7700X, just doing normal Windows things are super snappy. Love it.

 

[stan the caddy]

I recently built a pc with the ryzen 7600. 200 bucks and came with Starfield. Idle temp is about 38 and I can use the same board if I upgrade down the road. I wouldn't build anything with intel until the new cpus come out.

 

[mouser]

Been trying to get a used EVGA 3080ti FTW Ultra on EBay, but they’re consistently selling for $630-$700 after including shipping and sales tax.

At that price point it seems hard to justify over buying a new 4070 or 4070ti. My goal is running 1440p Ultra setting at consistent 60+ fps (1% rating) on the latest AAA releases. The game benchmarks suites for 1440p Ultra I’ve looked at have the 3080ti slightly ahead of the 4070 and slightly behind the 4070ti for fps. Want the Ray tracing and AI cores so AMD isn’t really a consideration.

Still bothers me all of these cards are only 12GB, would really prefer 16GB+ for a little more future proofing.

 

[PeteWorrell]

Been trying to get a used EVGA 3080ti FTW Ultra on EBay, but they’re consistently selling for $630-$700 after including shipping and sales tax.

At that price point it seems hard to justify over buying a new 4070 or 4070ti. My goal is running 1440p Ultra setting at consistent 60+ fps (1% rating) on the latest AAA releases. The game benchmarks suites for 1440p Ultra I’ve looked at have the 3080ti slightly ahead of the 4070 and slightly behind the 4070ti for fps. Want the Ray tracing and AI cores so AMD isn’t really a consideration.

Still bothers me all of these cards are only 12GB, would really prefer 16GB+ for a little more future proofing.

They don't want any future proofing. They learned from the longevity of the 1080Ti that people will hold on to a card like that for 5+ years so they have been cutting on features to get people to upgrade more often.

The exception is the 4090 which is a lot more costly than other cards so they don't mind throwing customers a bone on that one.

 

[mouser]

Disappointing but makes sense. Looking to upgrade my current 1070ti.

 

[PeteWorrell]

Disappointing but makes sense. Looking to upgrade my current 1070ti.

It's bad but the good news is that you are not paying a fortune like people who bought a 3070ti with only 8Gb of VRAM during the shortage. We can't forget that features like ray tracing take some extra VRAM so a low amount is quickly saturated.

 

[SolidSnakeUS]

Unless you really need DLSS (and I agree that it's f***ing awesome and personally, I have an RTX 3080), you could get a brand new 7900 XTX and would completely wipe the floor performance wise of the 3080 Ti for $940.

 

[stan the caddy]

Buying the 30 series is crazy to me. With nvidia you're paying a premium for the software. No DLSS 3 on the older stuff.

A couple of 4070 models got a price drop to 550 in the states.

 

[PK Cronin]

I've got an issue I'm hoping some of you fine folks might be able to help me with.

I recently upgraded to Windows 11 and have noticed that my audio is really low compared to where it was with Windows 10. I'm not able to hear sounds as easily on games like CS that I used to hear with ease (but it isn't exclusive to CS, it's all the time; listening to music, watching videos, talking on Discord). I've checked the volume mixer, loudness equalization, made sure my drivers are up to date, and anything else I could think of.

Has anyone experienced this or know of a fix?

Worth nothing that I didn't make any hardware upgrades when making the transition, everything is exactly as it was but with a newer OS.

 

[Osprey]

It appears to be a common problem with lots of guides to possibly fix it.

Google Search

 

[PK Cronin]

It appears to be a common problem with lots of guides to possibly fix it.

Google Search

I did a bunch of them, was hoping someone here had a fix they used themselves so I can just skip all the generic "run the audio troubleshooter" and get to the real solutions.

I saw one video suggesting an piece of software but haven't had time to try it yet (and I'm also skeptical of those sorts of downloads).

 

[Osprey]

I did a bunch of them, was hoping someone here had a fix they used themselves so I can just skip all the generic "run the audio troubleshooter" and get to the real solutions.

I hear ya. The guides with the generic fixes rarely work for me, too. I tend to have better luck with the search results that are posts on forums, so you may want to concentrate on those.

If I were you, I'd try uninstalling the sound card driver software, uninstalling the device from Device Manager, rebooting and letting Windows pick the best driver for you. If that works, I'd probably just leave it alone, but if you really want the sound card driver software, you could try installing it again.

 

[FriendlyGhost92]

I know this isn't necessarily the right thread for this, but it's the most applicable one I can find...

Anybody happen to know if the MacBooks with the Apple-made chips (M1/M2/M3) are any significant difference/improvement from the Intel-chipped MacBooks?

I ask mainly because I'm sick of every Macbook I buy having the battery go bad after a year and a half to two years, plus the casing itself heating up to the temperature of the sun until the fan kicks on and sounds like Hurricane Katrina rolling through because the RAM can't handle simple web browsing.

HFBoards has been notoriously bad at tanking the RAM on any computer I use, Mac or PC. I don't know if it's the ads or what, but it significantly effects RAM usage.

 

[Canes]

I know this isn't necessarily the right thread for this, but it's the most applicable one I can find...

Anybody happen to know if the MacBooks with the Apple-made chips (M1/M2/M3) are any significant difference/improvement from the Intel-chipped MacBooks?

I ask mainly because I'm sick of every Macbook I buy having the battery go bad after a year and a half to two years, plus the casing itself heating up to the temperature of the sun until the fan kicks on and sounds like Hurricane Katrina rolling through because the RAM can't handle simple web browsing.

HFBoards has been notoriously bad at tanking the RAM on any computer I use, Mac or PC. I don't know if it's the ads or what, but it significantly effects RAM usage.

Battery life is apparently much better with the M chips.

Also use a browser that supports uBlock Origin.