Blur Busters Forums

Eonds wrote: ↑

08 May 2022, 20:18

ZetaAlo wrote: ↑

08 May 2022, 19:40

Eonds wrote: ↑

08 May 2022, 01:38

ZetaAlo wrote: ↑

08 May 2022, 00:47

Ooh the broke crying pseudo intellectual is mad I can afford 4000 CL 14 ram and he can't, so he comes out with hostility and accusations with an extremely condescending tone, offering 0 helpful information, just spite. I know what I'm doing you clown. Lmao. I'm running it at 3733, CL14. I even ran it at 3800 with no errors for hours, but I couldn't get super tight timings. I'm aware of the FLCK limitations if you want a 1:1 ratio. It is of course, dependent on your silicon quality and I got a good one. Also, the ram timings effecting mouse lag isn't well known at all. You're simply mad. Because you're broke? Get your money up. Clown. Respond with actual helpful information next time instead of being mad and spiteful you're using a PC from 2010 and I wouldn't have to roast the fuck out of you like this

What are you talking about? If you don't run a 1:1 FCLK to MCLK ratio you get extreme latency and worse performance. Ram timings are obviously able to impact mouse feeling if you're reducing the latency of the DRAM. It's actually common sense but I know a lot of weird people will deny it even though it doesn't take a genius to draw that conclusion. I'm not broke by the way but that's fine ig. I don't have a PC from 2010 & no ones mad or spiteful. I don't want a back and forth but aren't you the one who bought an AMD CPU and talking about latency..... Anyway I hope I cleared that up. DRAM is at least worth 10% of total system latency. So if you're reducing the latency of the DRAM then................ you're reducing system latency. If you're saying the average person doesn't know then that's fine the average person doesn't know much anyway. If you're saying this as if it was some ground breaking discovery well it's not but good on you for being aware enough to realize it.

You were obviously being condescending in your first post. Come on. I'm saying it's a real issue and it's weird that it isn't mainstream. No shit lower latency ran will reduce system latency, especially on Ryzen which uses the infinity fabric to communicate with multiple processes. It's just weird this isn't talked about on YouTube or Techtubers etc. I guarantee if you contacted AMD/INTEL etc they would vehemently deny it being related aswell

Anyone who is directly involved with latency related work will confirm it. The topic is magnitudes more complex not only to understand but to explain. That's mostly why. As for the "do you think i'd buy X and X if I didn't know how it works". Yes I think you would & there's a good chance it's unstable & not only unstable but you bought a ryzen cpu anyway so that is a red flag. The red pill is that if you want low latency buy low latency hardware and stop caring about other metrics. If you're playing video games competitively low latency & smoothness is #1 priority nothing else. Buying a sub 5ghz capable cpu was the first mistake. Part 2 of the red pill is that 99.99999% of people have no idea what they're talking about + the only people who would tell you is people who specialize in specific areas who could care less about anything when they make 7+ digit $ a year.

Again, all I hear is pseudo intellectual, condescending dribble. From someone whose post history screams neuroticism, I guarantee I know more on this subject than you. This isn't Zen 2, . 5950X has a faster Cache Than Most Intel Processors, especially my previous 9900K, even at a lower clock. It also has more of it, and a higher IPC. I instantly felt a difference in snappiness and responsiveness the first day I booted it up. Core to Core latency on cores on the same CCX is also much lower than anything Intel has available. The only place where the 5950X falls short is DRAM/MEMORY latency. I.E infinity Fabric/Communication betweens cores on different CCXs. Clearly you don't have much knowledge about the subject even given your obsessive nature and the amount of time you've been on the forum. Anyways, my rams Stable, almost at 55NS latency in AIDA64 Aswell. My 9900K was at 54. I know you're mad, but...just get your money up.

dervu wrote: ↑

09 May 2022, 04:07

Too bad that guy on YT that did comparison input lag tests on Intel vs AMD did not make mouse movement latency moving from one point to another, so it would be obvious.
For just button click to on screen photon change there is not much difference that would matter for anyone.

However I have to admit that some issues related to consistency of mouse movement and game feel are truly hard to tell by 1000fps recordings while it is obvious to eye straight away.

Yeah, I notice for example, changing my TRFC even 50 or so units FROM SAY 360 to 310. Has a huge effect on general system latency AND mouse feel but doesn't show changes much in AIDA64. Webpages load near instantly, YouTube videos and other programs/apps aswell, the difference between 360 and 310 being no red buffer bar appears when I'm spam clicking on YouTube videos one after another. They literally load instantaneously. (TRC at 36 aswell) repeatable over and over and the opposite with say a 360 TRFC.

ZetaAlo wrote: ↑

09 May 2022, 08:25

dervu wrote: ↑

09 May 2022, 04:07

Too bad that guy on YT that did comparison input lag tests on Intel vs AMD did not make mouse movement latency moving from one point to another, so it would be obvious.
For just button click to on screen photon change there is not much difference that would matter for anyone.

However I have to admit that some issues related to consistency of mouse movement and game feel are truly hard to tell by 1000fps recordings while it is obvious to eye straight away.

Yeah, I notice for example, changing my TRFC even 50 or so units FROM SAY 360 to 310. Has a huge effect on general system latency AND mouse feel but doesn't show changes much in AIDA64. Webpages load near instantly, YouTube videos and other programs/apps aswell, the difference between 360 and 310 being no red buffer bar appears when I'm spam clicking on YouTube videos one after another. They literally load instantaneously. (TRC at 36 aswell) repeatable over and over and the opposite with say a 360 TRFC.

Can you reduce the values just like that, without any worries ? Or does that also depend on the current ram speed? The only thing I could find out now was that FCLK and MCLK must have identical values.

My tRFC value is 385 and Ram speed is 2400 Mhz. Can I just lower it and only this value ? Or do I have to adjust the values of TFAW and TRC as well ?

Unreazz wrote: ↑

09 May 2022, 09:08

ZetaAlo wrote: ↑

09 May 2022, 08:25

dervu wrote: ↑

09 May 2022, 04:07

Too bad that guy on YT that did comparison input lag tests on Intel vs AMD did not make mouse movement latency moving from one point to another, so it would be obvious.
For just button click to on screen photon change there is not much difference that would matter for anyone.

However I have to admit that some issues related to consistency of mouse movement and game feel are truly hard to tell by 1000fps recordings while it is obvious to eye straight away.

Yeah, I notice for example, changing my TRFC even 50 or so units FROM SAY 360 to 310. Has a huge effect on general system latency AND mouse feel but doesn't show changes much in AIDA64. Webpages load near instantly, YouTube videos and other programs/apps aswell, the difference between 360 and 310 being no red buffer bar appears when I'm spam clicking on YouTube videos one after another. They literally load instantaneously. (TRC at 36 aswell) repeatable over and over and the opposite with say a 360 TRFC.

Can you reduce the values just like that, without any worries ? Or does that also depend on the current ram speed? The only thing I could find out now was that FCLK and MCLK must have identical values.

My tRFC value is 385 and Ram speed is 2400 Mhz. Can I just lower it and only this value ? Or do I have to adjust the values of TFAW and TRC as well ?

As far as I know, FCLK and MCLK are the only things that must be identical or you get a performance penalty. You Don't have to do the same with timings, although to get the best performance out of your ram, you'd want to lower them all as much as possible. I.E you can lower TRFC to 330 and get better performance without lowering anything else...but if you lowered TRC too you'd get an even better performance boost.

Be careful though. TRFC and some other timings can cause some wonky and weird issues if they are too low/unstable. If I put TRFC too low, for example, My FPS drops massively whenever anything demanding is going on. Programs load slower, and some loading screens for easy to run games literally take like 5 minutes compared to like 10 seconds that they normally do. Lmao. Like in Destiny 2 I've had times where it took like 2 minutes after launching a destination for my ship to finally move and enter the "loading screen" (when it's normally like 3-5 seconds with Stable and Fast MEMORY OC)You can witness the instability and inconsistency real time. My sound Even goes out randomly sometimes with a "x" over my sound device like nothing's plugged in. It can even corrupt your OC in worse case over time.

ZetaAlo wrote: ↑

09 May 2022, 08:20

Eonds wrote: ↑

08 May 2022, 20:18

ZetaAlo wrote: ↑

08 May 2022, 19:40

Eonds wrote: ↑

08 May 2022, 01:38

What are you talking about? If you don't run a 1:1 FCLK to MCLK ratio you get extreme latency and worse performance. Ram timings are obviously able to impact mouse feeling if you're reducing the latency of the DRAM. It's actually common sense but I know a lot of weird people will deny it even though it doesn't take a genius to draw that conclusion. I'm not broke by the way but that's fine ig. I don't have a PC from 2010 & no ones mad or spiteful. I don't want a back and forth but aren't you the one who bought an AMD CPU and talking about latency..... Anyway I hope I cleared that up. DRAM is at least worth 10% of total system latency. So if you're reducing the latency of the DRAM then................ you're reducing system latency. If you're saying the average person doesn't know then that's fine the average person doesn't know much anyway. If you're saying this as if it was some ground breaking discovery well it's not but good on you for being aware enough to realize it.

You were obviously being condescending in your first post. Come on. I'm saying it's a real issue and it's weird that it isn't mainstream. No shit lower latency ran will reduce system latency, especially on Ryzen which uses the infinity fabric to communicate with multiple processes. It's just weird this isn't talked about on YouTube or Techtubers etc. I guarantee if you contacted AMD/INTEL etc they would vehemently deny it being related aswell

Anyone who is directly involved with latency related work will confirm it. The topic is magnitudes more complex not only to understand but to explain. That's mostly why. As for the "do you think i'd buy X and X if I didn't know how it works". Yes I think you would & there's a good chance it's unstable & not only unstable but you bought a ryzen cpu anyway so that is a red flag. The red pill is that if you want low latency buy low latency hardware and stop caring about other metrics. If you're playing video games competitively low latency & smoothness is #1 priority nothing else. Buying a sub 5ghz capable cpu was the first mistake. Part 2 of the red pill is that 99.99999% of people have no idea what they're talking about + the only people who would tell you is people who specialize in specific areas who could care less about anything when they make 7+ digit $ a year.

Again, all I hear is pseudo intellectual, condescending dribble. From someone whose post history screams neuroticism, I guarantee I know more on this subject than you. This isn't Zen 2, . 5950X has a faster Cache Than Most Intel Processors, especially my previous 9900K, even at a lower clock. It also has more of it, and a higher IPC. I instantly felt a difference in snappiness and responsiveness the first day I booted it up. Core to Core latency on cores on the same CCX is also much lower than anything Intel has available. The only place where the 5950X falls short is DRAM/MEMORY latency. I.E infinity Fabric/Communication betweens cores on different CCXs. Clearly you don't have much knowledge about the subject even given your obsessive nature and the amount of time you've been on the forum. Anyways, my rams Stable, almost at 55NS latency in AIDA64 Aswell. My 9900K was at 54. I know you're mad, but...just get your money up.

54ns on a 9900k is really bad. I get sub 43ns on a 3200 16-18-18-36 (dont care to post the rest) with a i7-10700k (4.8 ring 4.8 core). I already know everything you said.
[Paragraph containing Forum Rule Violation Deleted]
Just because you have a BIG l3 Cache that doesn't magically solve or compensate for terrible memory latency. My definition of bad may be different from yours. You're using clown emojis on a forum talking about DRAM latency. Trust me no one thinks your cool & you bought a sub 5ghz, terrible DRAM latency CPU & now you're coping and coming after me. People who have money and knowledge and also shoot for low latency & smoothness don't go for ryzen CPUS (i have a better system than yours). I don't know why you keep talking about money, using clown emojis, & insinuating that ryzen CPUS are somehow outperforming intel cpus (at the highest end) in latency. " The only part the 5950x falls short is DRAM latency" That's just not true there's many more niche things. You also say that casually like it's not important. Also the 5950x doesn't have better C2C latency than a good intel cpu. Since you claim I don't know anything about cpu architecture how did I know that the 5950x actually has worse overall C2C latency than my CPU (i7-10700k) ? Seems like it's you who doesn't know enough about the topic at hand.... I want to keep it civil and hopefully this post is approved so people can learn a little bit and think before they go directly attacking people.

ZetaAlo wrote: ↑

17 May 2022, 14:37

Eonds wrote: ↑

09 May 2022, 18:26

ZetaAlo wrote: ↑

09 May 2022, 08:20

Eonds wrote: ↑

08 May 2022, 20:18

Anyone who is directly involved with latency related work will confirm it. The topic is magnitudes more complex not only to understand but to explain. That's mostly why. As for the "do you think i'd buy X and X if I didn't know how it works". Yes I think you would & there's a good chance it's unstable & not only unstable but you bought a ryzen cpu anyway so that is a red flag. The red pill is that if you want low latency buy low latency hardware and stop caring about other metrics. If you're playing video games competitively low latency & smoothness is #1 priority nothing else. Buying a sub 5ghz capable cpu was the first mistake. Part 2 of the red pill is that 99.99999% of people have no idea what they're talking about + the only people who would tell you is people who specialize in specific areas who could care less about anything when they make 7+ digit $ a year.

Again, all I hear is pseudo intellectual, condescending dribble. From someone whose post history screams neuroticism, I guarantee I know more on this subject than you. This isn't Zen 2, . 5950X has a faster Cache Than Most Intel Processors, especially my previous 9900K, even at a lower clock. It also has more of it, and a higher IPC. I instantly felt a difference in snappiness and responsiveness the first day I booted it up. Core to Core latency on cores on the same CCX is also much lower than anything Intel has available. The only place where the 5950X falls short is DRAM/MEMORY latency. I.E infinity Fabric/Communication betweens cores on different CCXs. Clearly you don't have much knowledge about the subject even given your obsessive nature and the amount of time you've been on the forum. Anyways, my rams Stable, almost at 55NS latency in AIDA64 Aswell. My 9900K was at 54. I know you're mad, but...just get your money up.

54ns on a 9900k is really bad. I get sub 43ns on a 3200 16-18-18-36 (dont care to post the rest) with a i7-10700k (4.8 ring 4.8 core). I already know everything you said.
[forum rule violation]
[ForumRuleViolation]You're literally just projecting a bunch of random words you learned from your first psychology class. I don't care to argue about something so stupid. Highly knowledgeable people including myself don't use forums like this often due to bans/safe space/censoring/etc. Each of my posts has to be checked by them for it to be confirmed because I told someone they were delusional with their claims.[/ForumRuleViolation] Just because you have a BIG l3 Cache that doesn't magically solve or compensate for terrible memory latency. My definition of bad may be different from yours. You're using clown emojis on a forum talking about DRAM latency. Trust me no one thinks your cool & you bought a sub 5ghz, terrible DRAM latency CPU & now you're coping and coming after me. People who have money and knowledge and also shoot for low latency & smoothness don't go for ryzen CPUS (i have a better system than yours). I don't know why you keep talking about money, using clown emojis, & insinuating that ryzen CPUS are somehow outperforming intel cpus (at the highest end) in latency. " The only part the 5950x falls short is DRAM latency" That's just not true there's many more niche things. You also say that casually like it's not important. Also the 5950x doesn't have better C2C latency than a good intel cpu. Since you claim I don't know anything about cpu architecture how did I know that the 5950x actually has worse overall C2C latency than my CPU (i7-10700k) ? Seems like it's you who doesn't know enough about the topic at hand.... I want to keep it civil and hopefully this post is approved so people can learn a little bit and think before they go directly attacking people.

[forum rule violation]
[ForumRuleViolation]All i see is a clown with zero self awareness who is poor[/ForumRuleViolation], your system isn't better than mine, you're simply a clown with a god complex and a fragile ego, that's why the mere mention of my system specs and my 4000 cl14 ram made you seethe and cry and bark when i was simply asking if anyone else experienced my issue. Imagine having such a frail and fragile ego, you get insecure about me being able to afford better parts than you, and become hostile and condescending with your first response. Your first response was disgustingly condescending for no reason at all, pathetic. Lmao, go out and accomplish something in your life, fix that ego. Anyways, https://www.anandtech.com/show/16214/am ... x-tested/5 by core to core latency, i'm referring to cores on the same CCX, obviously. In that regard 5950X IS superior to Intel for the most part. THE ONLY major issue with current Ryzen architecture is, again, inter core latency between cores on different CCX's due to the fact That Ryzen still uses Infinity Fabric and That has a gigantic penalty on CCX-CCX communication/DRAM and certain other things that use IF. Ryzen has faster L3 cache, larger L3 cache, once it fixes it's infinity fabric latency issue, it'll be completely dominant. Also, we both saw the performance of the 5800X3D, still uses infinity fabric, still has a ram/memory penalty yet it poops all over even the best Intel gaming CPU in terms of latency/smoothness/frame times in MOST games. Which does show that cache size can compensate immensely for the infinity fabric latency. Again, get your money up and maybe you wouldn't be so insecure.

I have a 4000 c14 kit as well but okay. Again you're being full weird. It's not an issue, it's just how DRAM works.... I don't care that you can buy a ryzen cpu I promise. In competitive FPS games it doesn't out perform a high end intel cpu because games still access DRAM and those specific games care about memory latency a lot. It performs worse in CPU bound competitive FPS shooters which seems more relevant for this specific topic anyway. You keep talking about money for no reason. I genuinely find that so weird. I have high end equipment and a lot of experience. If you continue using (false) insults you might get banned/warned. Cache exists to compensate for DRAM's downfalls, but you can't print FAST ENOUGH cache to compensate for memory latency. Inter core latency overall is really important so let's not ignore THE FULL PICTURE because it's convenient. I could care less about new architecture anyway since I prefer lower latency designs like the ring architecture. My point is and was ryzen isn't lower latency anyway & of course DRAM directly has an impact on system latency. It's not new and if you understood how pc's work fully you wouldn't make a title like "5950X Ram Timings Effect mouse responsiveness massively??" as if that's somehow only impacting ONE ARCHITECTURE and not a universal thing. I don't care about intel vs amd argument. I use whatever has the best latency and performance which is still 10th gen intel in my opinion (assuming its fully maxed out). I know a lot about DRAM/CPU/PCB design and I'm telling you for a Fact that any amd cpu is not outperforming a maxed out 10th gen cpu in latency. I don't care that you bought a ryzen CPU, I'm strictly telling you that the purchase you made may not be the "lowest latency". Hopefully that cleared that up. If you believe that I'm wrong & that your purchase is somehow better that's okay. I whole heartedly don't care for the insults you're using.

Eonds you have really low Latency is it about hardware or maybe your OS settings?
what is your timer Resolution setting 0.5 or 1 ?

Eonds wrote: ↑

17 May 2022, 19:49

Fact that any amd cpu is not outperforming a maxed out 10th gen cpu in latency. I don't care that you bought a ryzen CPU, I'm strictly telling you that the purchase you made may not be the "lowest latency". Hopefully that cleared that up. If you believe that I'm wrong & that your purchase is somehow better that's okay. I whole heartedly don't care for the insults you're using.

Eonds/ZetaAlo, both of you have violated forum rules fighting each other.

Devil's Advocate: If you disable half the cores of a 5950X for a specific esports game -- then none of the cores exceed 20ns latency like it can on an i7-10700K. While both have quirks, some parity is achieved by just nerfing half of the cores temporarily. Tweakers sometimes use thread-affinity tricks to get those lowest-lag cores talking to each other in specific critical apps, to improve AMD performance scores.

So AMD<Intel vs AMD>Intel is sometimes a matter of unconventional left-field tweaking, but some of it is fair game...

There are other latency problems that can come from other things (e.g. the PCIe lane performance and whatever), so Intel and AMD can have different attributes that can be highlighted.

Yes, CPU-to-CPU latencies isn't everything (there's lots of problems in both Intel and AMD as we try to milk the last vestiges of Moore's Law), but one can use tweaking utilities to switch CPU profiles with fewer cores during esports and more cores during video rendering, etc.

Use a software utility for your 2-to-4-core esports game, and...

USB processing latency can be different problems that is some times unrelated to core-to-core latency, and sometimes fixable by just hopping onto a PCIe lane and using a USB3 card instead for your 8000hz mouse to stop worrying about AMD-vs-Intel chipset mouse quality differences.

An ugly workaround, but it's just a $25 workaround from Amazon/ebay, but it puts your mouse on a more express lane than many crappy motherboard USB implementations that are running on heavily-shared-overhead chipsets -- especially if the only device connected to your PCIe USB card is your 8000 Hz mouse.

One can still milk the CPU that they already have, with some simple common sense stuff.

Now if you wanted to disable more cores and isolate to only 8 cores because your esports game only uses 2-3 cores and mouse driver uses only 1 core, then go nuts and disable 75% of your CPU temporarily. The AMD numbers then look slightly better than Intel when you temporarily nerf your CPU in esports, especially when combined with an external PCIe USB3 8000Hz-capable card to bypass crappy on-chipset USB controllers. Much better esports parity happens at the end of the day.

No it doesn't always solve all problems. AMD can still be worse for certain things but so Intel can be worse for other things. You cherrypick something that can be worse. You got other things like RAM timings or other overheads. Neither CPU is best-in-all-class.

The thread topic is about USB mouse responsiveness, is it not? (Rheoretical).

Y'know, the USB problem is easy to fix. Step sideways, not backwards or forwards for this one. Bypass the chipset and get a PCIe USB card that's got excellent 8000Hz performance. Mouse drivers aren't RAM-latency bottlenecked.

There are ways to clears up a few weak links cheaply without an expensive motherboard replacement -- time-expensive and money-expensive for many people in this worldwide forum, as not everyone can easily buy Intel as easily/cheaply as AMD or vice-versa in their specific country, etc, etc, etc, etc, yadda, yadda, etc. The price differential playing field can tilt more heavily in certain countries. Or maybe a family got you the system, and you need to milk in a "what I can do" manner.

And then when exiting esports games, you can re-enable all your CPU cores to run your 32-core Threadripper software for speedy Adobe Premiere editing of your esports recordings.

The fence is not always greener on the other side, especially when cheap fixes exist for targeted problems like USB bugs.
(HINT: USB problems aren't caused by RAM latency problems. Look at the topic line, we're debating mouse responsiveness. Manufacturing arguments for the sake of bashing the other forum member isn't a good on-topic use of the thread subject. Focus on the cheap fix.)

The bottom line is stop violating forum rules, both of you
*Forum rule violations includes criticism about our moderating or about our rules
*Unapproved & approved rule-violating posts will still be reprimanded. Don't test the queue.

Thread pre-emptively closed.

</micdrop velocity="thrown" target="out-of-window">