FPS less after After Hours update

[Gaffa]

After starting the game, kill the 'SocialClubHelper' process via Task Manager. 

 

Rockstar added in checks which could be slowing things down. I believe today's patch is an attempt to fix this though, but try killing off that process once the game's started and it should help. 

[LunaStrky :3]

6 hours ago, Gaffa said:

After starting the game, kill the 'SocialClubHelper' process via Task Manager. 

 

Rockstar added in checks which could be slowing things down. I believe today's patch is an attempt to fix this though, but try killing off that process once the game's started and it should help. 

Dude, there are like 6 of them, wich one's the good one ._.'

[Hunter241]

11 hours ago, Gaffa said:

After starting the game, kill the 'SocialClubHelper' process via Task Manager. 

 

Rockstar added in checks which could be slowing things down. I believe today's patch is an attempt to fix this though, but try killing off that process once the game's started and it should help. 

 

That patch may have fixed it

Before, I was getting around 60fps on a single GPU, last night I was hitting near 100 on a 980ti @ 1440p

Gonna re-enable sli and see what I get, before GTAV used to get some positive scaling, hopefully that’s still the case

Edited October 17, 2018 by Hunter241

[DylRicho]

20 hours ago, fatdog said:

He can disable HT and get almost same fps with 100% CPU utilization. Because games performance never scaled well from number of CPU cores. Good engines can utilize more cores, but there is always a limit, unlike gpu, so please don't mix these two things. If you still not catching up, watch this https://m.youtube.com/watch?v=NVyo0p3Sd0E

 

There are numerous things wrong with that video.

 

1. GTA is an open-world game. That means with every single session, there's something different going on, and that affects the CPU and GPU figures. This is also present in the benchmark that comes with the game. The AI in the game makes it very CPU-intensive (more so at further draw distances), but the unpredictability in the AI spawning, interaction and animation makes the benchmark useless at best. It's even more useless when your benchmark consists of an uncontrolled route around the city, knocking people down (that will never react the same way twice), like in that video.
   
   In my opinion, Rockstar should have added an absolute static benchmark; one that still takes place using various modes of transport in the game, but with traffic and pedestrians fixed with regards to spawning and animations. The complete randomness makes the numbers it spits out meaningless.
   
    
2. The guy falsely claims that the CPUs are clocked at 3.80 GHz, but that's not the case. Under multi-threaded workloads, it's 3.60 GHz on all cores. At first, I gave him the benefit of the doubt because I saw the CPU clocked at 3.80 GHz with the "Core i3" test, but further along in the video, you can see that all he's done is restrict the active core count, and left the frequency and L3 cache unchanged. At best, you're looking at results, clocked between 3.40 and 3.60 GHz.
   
   I also don't agree with lumbering Skylake models in with his strange idea of a benchmark. It's a Haswell-based Xeon, with Haswell IPC. Sure, you can claim they might be similar, and they probably will, but I didn't get that from the video. The only thing worse than this, would be if he was also trying to convey that AMD's Zen processors were similar in performance. They are completely different architectures, and that would be nothing short of unacceptable.
   
   As it is, you can't draw conclusions for newer processors, based on an older one, no matter how close the IPC is. Rough approximations, sure. But what I find more annoying about this, is that the core frequencies and L3 cache don't match up with his claimed processor models. I get it. There's only so much you can do with a Xeon, but in that case he should have never made this video in the first place.
   
    
3. What the video shows, is pretty much what has been discussed. The results going from the "Core i5" to "Core i7" don't really change at all. The extra threads of the Core i7 would get a little more out of each core as it can be fully saturated, but the CPU load figures are not pegged at 100%, which would be the sign of a bottleneck. Even in the "Core i5" test, they weren't pegged.
   
   The game has a frame limit cap of 188 FPS. You can achieve this with an overclocked quad-core Core i3, and so any additional cores (e.g. moving up to an i9-9900K for the cores and not the frequency, or a Ryzen 7 2700X) will not be put to good use. I hope it changes for the future, and I'm sure it will, but right now I will once again emphasize that GTA likes instruction throughput, but up to a certain point, the additional throughput does absolutely nothing. An example would be moving from an i7-7700K up to an i7-8700K; the 50% extra threads net you just 8% more frames (at identical frequencies).

Edited October 17, 2018 by DylRicho
Clarified a few things.

[fatdog]

3 hours ago, DylRicho said:

The results going from the "Core i5" to "Core i7" don't really change at all.

This is basically why I linked that video. No benefit from more threads\cores, due to engine limitations. Anyway, I've made my own video, with detailed cores load, that should further explain my point https://youtu.be/SbrOf8g0fRc

There you can see full utilization of gta.exe process, I have 6 active cores total, so each one goes for 15% roughly. Most important part is on the threads tab, there are always one or two maxed out threads - and this is your bottleneck, a single maxed out core, that determines general performance. The game engine is simply unbalanced, poorly designed sh*t, thats why it needs CPU with faster cores. 

 

 

 

[DylRicho]

2 hours ago, fatdog said:

The game engine is simply unbalanced, poorly designed sh*t, thats why it needs CPU with faster cores. 

 

But that's what we've all been saying? 

Edited October 17, 2018 by DylRicho
Removed useless whitespace.

[fatdog]

2 minutes ago, DylRicho said:

 

But that's what we've all been saying? 

Nah, you denying CPU bottleneck.

[DylRicho]

37 minutes ago, fatdog said:

Nah, you denying CPU bottleneck.

 

No, we both agree that there is a bottleneck. However, the bottleneck is in RAGE, and not in a quad-core processor.

 

You have literally just agreed with me that there's no point upgrading beyond a quad-core Core i7 because frame rate doesn't increase, and yet you were telling @TheRealDadBeard earlier in the thread that the quad-core Core i7 was hindering performance.

 

Don't change your view because you have been corrected.

[fatdog]

26 minutes ago, DylRicho said:

No, we both agree that there is a bottleneck. However, the bottleneck is in RAGE, and not in a quad-core processor.

Oh my god, you ignorant clown. Bottleneck is in slow single core performance and I was telling the other guy to get something with better IPC and clock, and I still do.

[DylRicho]

10 minutes ago, fatdog said:

Oh my god, you ignorant clown. Bottleneck is in slow single core performance and I was telling the other guy to get something with better IPC and clock, and I still do.

 

Ignorant? Maybe you're just arrogant? 

 

Something with better IPC than Haswell? There is nothing in this case. Clocks, sure. But a 4790K can definitely be overclocked to the point where buying a newer processor is pointless.

[fatdog]

12 minutes ago, DylRicho said:

Something with better IPC than Haswell?

Yes, intel significantly improved it since 6th gen, especially combined with ddr4 ram.

[DylRicho]

3 minutes ago, fatdog said:

Yes, intel significantly improved it since 6th gen, especially combined with ddr4 ram.

 

Not for this game.

[fatdog]

5 minutes ago, DylRicho said:

Not for this game.

Well, you're wrong, ask any owner of these.

[DylRicho]

Just now, fatdog said:

Well, you're wrong, ask any owner of these.

 

I'm not wrong. You're under the common misconception that the IPC figures touted apply to absolutely every single workload sequence. That isn't true. It depends on the game engine in question, but generally IPC matters less in gaming than in other workloads such as video encoding (on the CPU) or rendering. It still makes a difference, but extra frequency will net you better results.

 

It's completely asinine to think that every single sequence of instructions a processor executes benefits to the same degree that is touted in marketing material. Haswell to Skylake in a rendering benchmark like Cinebench is demonstrated to be approximately 10.1%, but in gaming, and more specifically in Grand Theft Auto V, it's demonstrated to be 1%. Do you know how many frames that translates into? Half a frame. Not even one frame.

 

You're perfectly justified in suggesting a processor with higher frequency, but you seem to be forgetting that an i7-4790K can overclock just as well as an i7-6700K can. Why would you bother spending hundreds of dollars on another processor for the same frame rate? It makes no sense whatsoever.

[fatdog]

1 minute ago, DylRicho said:

but in gaming, and more specifically in Grand Theft Auto V, it's demonstrated to be 1%

Lmao, I call it bs. I jumped from 4770 to ryzen 2600x with 3200 mhz ram with significant improvement as less stutters and higher average framerate. Ryzen is bad for gta, but I guess still better than haswell. What is your comp, care to compare fps with me in a same session?

[DylRicho]

3 minutes ago, fatdog said:

Lmao, I call it bs. I jumped from 4770 to ryzen 2600x with 3200 mhz ram with significant improvement as less stutters and higher average framerate. Ryzen is bad for gta, but I guess still better than haswell. What is your comp, care to compare fps with me in a same session?

 

Ryzen is somewhat of a misnomer because the cores are all situated on top of a piece of silicon with data links going to each of the complexes. Those data links are directly tied to the frequency of the system memory you're using in a 1:1 ratio, which would be 1,600 MHz for your system. As a result, Ryzen loves faster memory and better timings in scenarios where Intel doesn't give a s**t. Intel is still using a ring bus between all cores.

 

I'm not using Intel or Ryzen, so my system's results aren't going to be useful.

 

The extra two cores in the 2600X demonstrate an increase of less than an extra frame (0.8%), inclusive of the frequency differences because these results are not normalized to one frequency like the others were. In frame times, the 2600X does demonstrate more fluid performance, spending just over half the time of the 6700K below 60 fps.

 

However, the 7700K performs better than the 2600X, implying that the extra two cores help very little. And, indeed, if you take the frame rates of the 7700K and 8700K and find the difference, you're met with 6.6%; very close to the 8% I provided earlier in the thread. That's atrocious for a 50% core increase.

[Geisterfaust]

1 hour ago, fatdog said:

Well, you're wrong, ask any owner of these.

I have both 4790k @4.6 GHz and since the last week 8700k at stock because I only use Noctua NH-12US and this beast asks for at least U14S or better D15. Guess what? At 1080p using the same Asus Strix 1080 Ti OC I'm getting almost the same drops while simultaneous sales are happening in the session. And speaking of your comparison, there's more difference between 4770 and 4790k at stock than between 4790k and 6700k in single core performace,  4790k is still rocking hard for single core performance (14% difference between 8700k). So yeah, I own these 2 systems and as I said earlier, full sessions with multiple sales will cripple anything that's on the market. There's no CPU to handle that and that's bad Online optimisation here. Single player works just fine even on 4690k, used to own this one too back in '16. 

 

1 hour ago, fatdog said:

Ryzen is bad for gta, but I guess still better than haswel

QC speed is 7% better in comparison to 4790k. Your gains are because you have moved from 4770 which is really slow: +11% SC and 24% QC. However, comparing with 4790k (at stock!) the difference almost disappears: +7% QC. Considering that I'm running 4790k at 4.6 stable, there should be no discussion. DDR3 vs DDR 4 in gaming won't do much difference at all, various tests proved that. 

 

 

Edited October 17, 2018 by hei3enberg

[DylRicho]

14 minutes ago, hei3enberg said:

QC speed is 7% better in comparison to 4790k. Your gains are because you have moved from 4770 which is really slow. 

 

@fatdog

 

This will be why the 2600X is getting better frame times. Good catch, @hei3enberg. I completely forgot about that aspect. 

 

Edited October 17, 2018 by DylRicho

[Geisterfaust]

20 minutes ago, DylRicho said:

This will be why the 2600X is getting better frame times

It should, yeah. 4770 is a very bad Haswell while 4790k is an excellent Haswell refresh: not only better thermals, but even disregarding it's a K, the stock is 4.0 with 4.5 boost against 3.6/ 3.9 boost at 4770. 

[fatdog]

6 hours ago, hei3enberg said:

Guess what? At 1080p using the same Asus Strix 1080 Ti OC I'm getting almost the same drops while simultaneous sales are happening in the session.

Weird, there is should be at least 20% gain from ipc and ddr4, you sure you kept the same settings, especially for view distance? Anyway that guy with locked 4790 could've use an upgrade.

[MajorVienna]

removed etc

Edited October 18, 2018 by MajorVienna

[DylRicho]

4 hours ago, fatdog said:

Weird, there is should be at least 20% gain from ipc and ddr4, you sure you kept the same settings, especially for view distance? Anyway that guy with locked 4790 could've use an upgrade.

 

What? I give up. 

[Geisterfaust]

10 hours ago, fatdog said:

Weird, there is should be at least 20% gain from ipc and ddr4, you sure you kept the same settings, especially for view distance? Anyway that guy with locked 4790 could've use an upgrade.

Well, with unlocked FPS there's more with 8700k than with 4790k, especially if you're going over 100, sure. But the drops are caused by bad coded sales and the drops are absolutely the same: 5-15 fps with the same settings. I can't imagine how cartel lobbies are doing on consoles when 3-4 orgs are selling at the same time. 

[fatdog]

12 minutes ago, hei3enberg said:

But the drops are caused by bad coded sales and the drops are absolutely the same: 5-15 fps with the same settings.

Isn't it better to drop from 80 fps other than 60 though?

[Geisterfaust]

10 minutes ago, fatdog said:

Isn't it better to drop from 80 fps other than 60 though?

Unless you're playing at 4k 60...

[fatdog]

1 minute ago, hei3enberg said:

Unless you're playing at 4k 60...

You saying haswell never drops you under 60? 

[Geisterfaust]

2 minutes ago, fatdog said:

You saying haswell never drops you under 60? 

I'm saying that 4790k @ 4.6 GHz produces the same bloody drops from 60 on 4k as 8700k. Same story with 1080p. With the same Asus ROG 1080 Ti OC @ stock OC. And that is annoying as hell, because I would like to play in 4k and I don't own 100+ Hz displays. I'm particularly describing one scenario: multiple sales in full session. It's broken. 

Edited October 18, 2018 by hei3enberg

[DylRicho]

On 9/28/2018 at 1:54 AM, TheRealDadBeard said:

Same for me. I tried everything but when the lobby fills up I get massive drops. Even lowering graphics settings doesn't help. Gladly I will update next month to a new GPU. They did something to probably makes the lights nicer and stuff and overdone it.

 

By the way, I too am getting useless performance whenever I try to interact with the goods. In fact, my game crashes.

 

Now I've figured out what it is. It's these stupid business battles. I'm staying away from them until they fix it.

Edited October 19, 2018 by DylRicho
Added link.

[fatdog]

5 hours ago, DylRicho said:

 

By the way, I too am getting useless performance whenever I try to interact with the goods. In fact, my game crashes.

 

Now I've figured out what it is. It's these stupid business battles. I'm staying away from them until they fix it.

Turns out you have really potato comp, give yourself a favor and upgrade it to 2200g at least?

[TheRealDadBeard]

9 hours ago, fatdog said:

Turns out you have really potato comp, give yourself a favor and upgrade it to 2200g at least?

Even better, do everyone a favor and just shut up and being annoying, Thanks on behalf of everyone.