Nebell
2[H]4U
- Joined
- Jul 20, 2015
- Messages
- 2,383
I think I still had my K6-2 back then.
![[H]ard|Forum](/styles/hardforum/xenforo/logo_dark.png){kind=logo}
Ryzen 3000 Processors Listed on Russian Retail Site
I think I still had my K6-2 back then.
NWRMidnight
Limp Gawd
- Joined
- Oct 23, 2010
- Messages
- 352
LOL!!
You are damn straight I was looking at the 2700x numbers, because the 2700x has a higher IPC than the 2600x, which is what this whole argument is about. Also, the ONLY game that utilizes more than 4 cores / 8 threads at once is ashes, the others do not, which means the extra cores mean nothing when it comes to the IPC gaming performance. I also ignored the 2600x results because most sites do not show the 2600x under performing in comparison to the 2700x with the performance between the two having only a couple fps in most games which is due to the IPC difference (and that is at stock speeds) . This article was also published 1 week after the release of the Zen+, using a motherboard that had temperature and throttling issues that was corrected with a later bios update. You also have to take into account for the memory settings. Yes they used the Exreme memory profiles, which is an INTEL specification, and AMD Ryzen Motherboards do not translate the profiles accurately, which is why it is recommended to manually input all the memory settings to get the proper and optimal performance. Also the Asrock X470 on release had issues with XMS profiles and not running at full speeds, it wasn't really an issue per say, but was related to XMP profile translations and setting that had to be manually set. That is why you need to take all benchmarks into account, not just the gaming to see that there is only a 2% to 5% in IPC difference between the Zen+ and Intel's offering.
If you really want to get into the nuts and bolts of things, Was the 2600x getting throttled due to the overclocking as it is a lower binned chip than the 2700x? Hence why it's default speeds are lower (seems like it was being throttled since there is such a large margin between the 2700x and the 2600x), which there shouldn't be in almost all games since they do not utilize more than 4 cores. IF you go look at all the benchmarks with the 9700K, you will see that the 8700k beats it in some games, and they have identical IPC, but yet, the 9700k has more cores. Go figure.
Point is, there are so many influences when it comes to gaming benchmarks, that they cannot be used as a indication of IPC performance.
Last edited:
Vulkan actually changes that somewhat, at least under Linux. Using htop I can see that games coded using the Vulkan API actually make use of all cores fairly evenly, and I'm running 24C/12T.
Perhaps hyperthreading plays more of a role in certain games than most assume?
Last edited:
legcramp
[H]F Junkie
- Joined
- Aug 16, 2004
- Messages
- 12,401
Does the 1800x have a higher IPC than the 1600x also? Seems like more than one game were using the extra cores. Also why is there a difference in IPC between 2700x and 2600x besides the L2 cache that shows such a huge boost for the 2700x?
My Ryzen 1600 system when it came out running B-die memory worked fine with XMP enabled out of the box, CPU was soldered and ran super cool compared to my 8700K system. I don't see how a legit site like techspot would have any of the overheating/throttling issues you're claiming .. unless you meant some of the intel chips were throttling if they weren't delidded? lol
9700K vs. 8700K is apples and oranges... 8 cores no HT vs 6 core / 12 threads... ?
Point is I am just copy and pasting whatever the article said and today I learned the 2700x has a way faster IPC than the 2600x.
Last edited:
dgz
Supreme [H]ardness
- Joined
- Feb 15, 2010
- Messages
- 5,838
I'll crap myself if 3800 ends up an 8 core lol.
thebufenator
[H]ard|Gawd
- Joined
- Dec 8, 2004
- Messages
- 1,383
Is that why my 2700x sits with all cores at a tad under 4.1 when playing modern games? Older games run with ~4 cores boosted higher and the other cores with a lighter load in the ~3.5 range.
NWRMidnight
Limp Gawd
- Joined
- Oct 23, 2010
- Messages
- 352
First, if you go look at techspots review of the 2600x and 2700x, their numbers are different then what they show in IPC/4.0Ghz comparison, in some cases they have both the 2700x and 2600x overclocked even higher yet getting worse performance numbers, and they are getting better performance at stock speeds in some games, which shows there is to much fluctuation in gaming benchmarks to make a legitimate determination (of course they are also all done 1 week after release of the Zen+). As for the 1800x, yes it has higher IPC than the 1600x. As for your memory running right out of the box, that was't the case for many people, there where some lucky ones, but not many. The first generation of Ryzen had issues with ram speeds for months after release, and in some cases still do. Also, showing running at 3200Mhz does not mean it is running at the same optimal settings as it does in an Intel board due to missing settings that the XMP do not have that are in the AMD specification, again, why it is recommended to set all ram settings manually.
As for the 9700k having NO HT, you are correct, I actually meant the 9900k.. sorry.
NWRMidnight
Limp Gawd
- Joined
- Oct 23, 2010
- Messages
- 352
just because the speed kicks up to 4.1 ghz does not mean that it is utilizing more than 4 cores. If you actually watch your per core usage, you will see them all jump all over the place, with a few having a higher load than others. Also, the memory controller shows it puts a load on the CPU (it really doesn't put much, but because of how the Ryzen memory controller is designed, it appears to). To show this, test your memory for stability and watch all cores go to 100%, even though the cores do not really have a full load on them, which is backed up by cpu temperature, as it changes very little during a memory stability test. If there truly was a full load on the CPU during a memory test, the CPU temperatures would also increase substantially, which they don't. Modern games use substantially more memory than older games. Hence why it appears you are using more cores, when you truly are not. Also, old games vs newer games are handled differently by the OS and how it allocates CPU usage.
Last edited:
schoolslave
[H]ard|Gawd
- Joined
- Dec 7, 2010
- Messages
- 1,293
No -- please explain how the OS is supposed to differentiate between an "old" and "new" game? All things equal, the OS treats a game just like any other process.
NWRMidnight
Limp Gawd
- Joined
- Oct 23, 2010
- Messages
- 352
Uh, yes, it is how older games passes on cpu calls to the OS, vs modern games. Hence why some older games don't work with new OS's as they try to use pathways or cpu calls that are no longer supported in a new OS, as well as calls to the gpu If that wasn't the case, then every game would utilize all cores, and it wouldn't be game specific, as all modern OS support multi cores, if what you say is true and the OS treats all applications the same. Actually most modern games have direct access to the cpu, but the OS still determines the "cpu time", meaning it still controls how many cpu cycles each application running gets, with the main application (games) getting the majority. The OS also distrubutes the work load to each core. That is why you see the usage bounce around on various cores. Specially more so in Windows 10 due to how it distributes tbe work load via it's schedular.
Last edited:
legcramp
[H]F Junkie
- Joined
- Aug 16, 2004
- Messages
- 12,401
How about a link to the benchmarks? And also, according to those 4.0Ghz comparison graphs, you're saying the 2700x has a 10% or so IPC advantage over the 2600x looking at those numbers and those higher values has nothing to do with the 2700x having more cores?
NWRMidnight
Limp Gawd
- Joined
- Oct 23, 2010
- Messages
- 352
Since you are to lazy to look for them :
https://www.techspot.com/review/1613-amd-ryzen-2700x-2600x/page3.html
Now go read what i said, and the reason i ignored the 2600x results, as they are not accurate, as the reveiw and benchmarks i just linked you show both the 2600x and the 2700x very close with the 2700x being slightly ahead of the 2600x due to the better ipc, and nothing to do with core count, except in ashes. This is at stock speeds (2600x is slower), and 4.1 Ghz and 4.2 Ghz over clocked. But yet, for some reason, the 2600x all of a sudden drops significantly in the 4Ghz comparison vs the 2700x. Yet, both articles are by the same place, but as an example, in ass creed they show the 2600x's performance being 9% less for the 4 Ghz comparison compared to itself in the other article. So.. yea..
Last edited:
legcramp
[H]F Junkie
- Joined
- Aug 16, 2004
- Messages
- 12,401
The only benchmark that shows any variation is the Assassin's Creed Ultra setting one and they mentioned GPU limited many times which is why the 2700x which was running 100Mhz higher than the 2600x at stock or overclocked has a slight lead.
Oh in case you want to take the GPU out of the equation even more, they conveniently have the BFV CPU benchmark using a 2080 Ti! Which shows once again a significant performance drop for the 2600X compared to the 2700X. Oh and if you want to compare the IPC with Intel since this isn't a clock-for-clock article, you can use the i5 8400 in this link which only boosts to 4.0Ghz which is 200-300Mhz lower than the Ryzen offerings.
https://www.techspot.com/review/1754-battlefield-5-cpu-multiplayer-bench/
NWRMidnight
Limp Gawd
- Joined
- Oct 23, 2010
- Messages
- 352
You keep moving the goal posts. There is only a 3 to 5% difference in IPC between AMD and Intel at identcal clock speeds. I already showed you one discrepency, i was also showing why you look at the 2700x when comparing IPC between AMD and Intel.
BTW the GPU bottle neck was for the 8700k, it was not bottlenecking the 2600x or the 2700x. As for your 2080ti reference, you will get a performance drop, as you have 2 extra cores that the 2080ti can offload some of it's work to on the 2700x, which all GPU's do, this is not the game utilizing the extra cores, but a monster GPU. Being cpu limited vs the gpu is no demonstration of IPC.
Last edited:
legcramp
[H]F Junkie
- Joined
- Aug 16, 2004
- Messages
- 12,401
Moving the goal post? I posted some benchmarks and you just change the subject again. So the i5 8400 which is running at a clock deficit is beating the 2700x yet you claim 3-5% difference in IPC?
I mean I am not the one making stuff up lol
And now you're claiming the 2700x has two extra cores the GPU can offload work to yet you claim earlier the 2700x is beating the 2600x not because of the extra cores but because it has a lot higher IPC... okay... LOL
I mean I am not the one making stuff up lol
And now you're claiming the 2700x has two extra cores the GPU can offload work to yet you claim earlier the 2700x is beating the 2600x not because of the extra cores but because it has a lot higher IPC... okay... LOL
Last edited:
NWRMidnight
Limp Gawd
- Joined
- Oct 23, 2010
- Messages
- 352
Here go read:
https://www.quora.com/How-does-the-graphics-card-work-with-the-cpu
First, i have already shown you the discrepancies with techspot. I never moved the goal posts but you don't want to use cinebench to determine single core performance, which directly measure IPC, and only want to use game performance. I explained why you use the 2700x for determining IPC between AMD and Intel, I used references using identical clock speeds to demonstrate it using only games, but you focused on the 2600x... Which is your mistake.
Now the i5 8400 is the same generation as the 8700k, which means it has the same IPC as the 8700k (as this is how Intel rolls, as the main difference between there chips of the same generation is core count, clock speed and HT), it boosts all cores to 3.8 Ghz, the 2700x, depending on cooling and tempurature will boost all cores to 3.9 to 4.1 Ghz.. if it is the stock cooler, most likely it boosts to 3.9 to 4Ghz in Battlefield one (dependent on the correct settings in bios) if the settings are wrong, or not enabled. no boost, just 3.7 Ghz. In otherwords there is really no clock deficit between the i5 8400 and the 2700x. One had HT or equivelent, one does not. So the 7fps (5%) difference is due to IPC as we really don't know to what extent HT effects BF1. That is 5% in one game, but thank you for picking out an example that supports that there is only a 5% differnce between AMD and Intel.
What's funny, in your desperate attempt to prove me wrong, you ignored the very fact that the 9900k is only 15% faster than the 2700x.. yet it runs nearly 25% higher clock speed than the 2700x, which says what about it's IPC in comparison to the 2700x to only be ahead by 15%.
Last edited:
legcramp
[H]F Junkie
- Joined
- Aug 16, 2004
- Messages
- 12,401
Yeah and it also has two extra cores which gives it a very generous boost over the 2600x, wouldn't it be more fair to compare the 8400 to 2600x since the monster GPU in this game is using the extra two cores of the 2700x? Or is it just the superior IPC of the 2700x giving it the lead or is the 2600x still throttling due to these benchmarks being run right after Ryzen release (The article came out in November 2018).
Yup, it's just one game and doesn't prove anything at all, just the gap is bigger than 3-5% no? lol
Do you fail to recognize the 2700x is handily beating the 2600x because it has two more cores?
Nice edits btw
Yup, it's just one game and doesn't prove anything at all, just the gap is bigger than 3-5% no? lol
Do you fail to recognize the 2700x is handily beating the 2600x because it has two more cores?
Nice edits btw
Last edited:
legcramp
[H]F Junkie
- Joined
- Aug 16, 2004
- Messages
- 12,401
I don't need to prove anything with the 9900K, it maxes out the 2080Ti at those settings which is why the 8600K range and up have very close numbers.
~2 fps difference in minimums lol
Not desperate at all, just find it funny we have AMD shills that are more delusional than the Intel ones. At least the Intel ones I've chatted with in the past don't move the goal post or make excuses for everything (moving subject to 2700x vs 9900K now aren't we? LOL).
I posted a graph up there for you to look at.. I bet you're going to reply with some BS though or change the subject
.BRB playing cinebench trying to unlock that crate box so I don't have to spend money on it, pretty sure the subject was gaming IPC but deflect more k.
Last edited:
NWRMidnight
Limp Gawd
- Joined
- Oct 23, 2010
- Messages
- 352
Sorry, i am on my phone, so posting, corrections are a little more difficult and time consuming.
The problem is, you fail to understand that it not the game that is creating the difference. The game is not using the extra cores. The 2080ti is throwing calculations at the 2600x that it can't handle fast enough, where as the 2700x has the extra cores and higher IPC to offload the workload of the 2080ti vs the 2600x. You can't deliberately bottleneck a cpu and then start saying it underperforms.
Not at all. NO game is that low level. Old Modern or in between.
I'm not sure why people think game developers are super coders. Outside of the few stand outs in the industry game developers are not super coders that dream in assembler. CPU affinity and process priority are all the domain of the OS.... even on consoles like PS4 and mobile devices like Android and IOS.
Developers code to System APIs. The OS CPU scheduler has the ball... always. Can you thread your code... yes, but the scheduler will decide if the system has the cores to do that... or not. No one is programming this bit for core 0 this bit for core 1 ect... that isn't how it works. You can thread your code so the cpu scheduler doesn't have to, and that is the fastest way to do things. But that has performance penalties as well if you over thread your code and the cpu scheduler has to wait for the next cycle.
The reason Game developers don't thread their code so the cpu scheduler goes yes look at this lovely 8 even bits of work is simple. For anyone that doesn't actually have 8 cores or threads.... the performance will nose dive. Cause the cpu scheduler will have to either recombine the work causing a bottle neck and a regression in performance vs un threaded code, or simply que the work up and take twice as many cycles causing even worse performance degradation.
So if your a game developer even today you have to be realistic and say Both consoles, have 4 cores. The majority of PC users still have 4-8 cores. So we can't create 10 thread paths... cause performance will completely tank for 90% of our customers. The other 10% will see mild improvements.... and only mild cause realistically even though their cpu schedulers may deal better with the better threaded code the preformance bump isn't going to be drastic unless those CPUs are really hitting the wall on 4 cores. (which non but the oldest 4-8 core chips are even with the latest game engines)
At the end of the day... game developers like all other software developers are mostly at the mercy of the cpu scheduler. If its well coded lean and mean things are going to run smooth, and yes well threaded code will run faster, and unthreaded code will still run very very well. On the flip side if you have a terrible OS developer who isn't even intelligent enough to keep an eye on performance when they dirty hack things... chances are multi threaded things will run like Garbage.
With the issues MS has been having with their cpu scheduler on high core chips... its more evidence that they would be better off releasing a windows 11 with a linux core system and a windows DE. They can keep DX and windows UI APIs closed... just remove MS monkeys from kernel / cpu / disk scheduler work.
Choosing Intel vs and in gnu limited gaming is a waste of time. Budget out a CPU in dollars and what you come up with price to performance? Unlimited budget and gaming only - sure Intel. Every other situation AMD is typically a better choice currently.
Running steam /tf2 (source game severely CPUs limited, threads to 2 CPUs default) I was up bottlenecking with R9290 in sli. Created a custom steam launch using cmd.exe and a cpu affinity mask to map steam.exe to the real cores (non-HT). One setting in the source engine to use in conjunction is mat_queue_mode "2" TF2 now threads on my six real cores and the bottleneck is gone. Same process could be used for any source engine game. It may help with other steam games but I haven't analyzed the fps on those yet.
edit: If anyone wants a howto, just ask and I can post in the game forum
Running steam /tf2 (source game severely CPUs limited, threads to 2 CPUs default) I was up bottlenecking with R9290 in sli. Created a custom steam launch using cmd.exe and a cpu affinity mask to map steam.exe to the real cores (non-HT). One setting in the source engine to use in conjunction is mat_queue_mode "2" TF2 now threads on my six real cores and the bottleneck is gone. Same process could be used for any source engine game. It may help with other steam games but I haven't analyzed the fps on those yet.
edit: If anyone wants a howto, just ask and I can post in the game forum
Last edited:
legcramp
[H]F Junkie
- Joined
- Aug 16, 2004
- Messages
- 12,401
Okay, so why did you conclude 8400 vs 2700x was a fair comparison and said it proves your 5% argument then when in fact the 2700x has two more cores to work with as well as threads?
NWRMidnight
Limp Gawd
- Joined
- Oct 23, 2010
- Messages
- 352
Hello, because we are talking about IPC. The i5 8400 does not have the issues the 2600x because it has 6% to 7% more IPC than the 2600x (remember the 2700x has better IPC than the 2600x), and also boosts higher on all cores than the 2600x, giving it's ability to handle the 2080ti much better. Each core on the i5 8400 has a higher IPC, multiply that by 6.. do the math. the 2700x is 5% lower in IPC, per CORE, do the math, (keep in mind the game is not utilizing all the cores on the 2700x, all though you keep implying that it does) You also disregarded my comment about threads. We do not know how battlefield 1 handles threads, as MOST games do not utilize threads. Hence, why it is recommended if you get stuttering or slow down in games to DISABLE HT. you lose NO gaming performance in 99% of the games out there when it is disabled, but can actually gain performance. That is why the I5 has been recommended as one of the top gaming CPU's if that is all you do with your machine, because HT has little or no impact on gaming.
You keep talking about the extra cores when gaming.. THE GAME does not use all the cores. The 2 extra cores are being utilized by the GPU's calculations and back ground applications, not the game. You are so focused on the cores, that most games don't use. (research it)..
ManofGod
[H]F Junkie
- Joined
- Oct 4, 2007
- Messages
- 12,864
The real issue is having this fixed without breaking backwards compatibility.
Oops: I thought I was in the 2990WX thread. Oh well, looking forward to the new Ryzen and GPU's from AMD, even though I will not be updating for a while.
Oops: I thought I was in the 2990WX thread.
Oh well, looking forward to the new Ryzen and GPU's from AMD, even though I will not be updating for a while.
legcramp
[H]F Junkie
- Joined
- Aug 16, 2004
- Messages
- 12,401
All I see is you making up crap in order to justify comparing the 2700X to the i5 8400, which the 8400 wins too mind you with a 100mhz worst case scenario deficit with two less cores.
Oh and the 7600K with no HT gets murdered in those charts.
Happy Holidays!
NWRMidnight
Limp Gawd
- Joined
- Oct 23, 2010
- Messages
- 352
I am not making things up. You just lack the knowledge to understand and keep fixating on those 2 extra cores that are not being used by the game, all while igoring rhe IPC differences.
Here, these benchmarks completely remove any load on the cpu caused by the gpu and strictly testing the cpu and it demonstrate the core count has no impact . . Take note at the BF 1 results, specifically the 2700x and the 2600x (heck look at all the games, notice how close the 2600x and the 2700x are... Oh wait, impossible because the 2700x has those 2 extra cores you are so fixated on.)
It also shows that HT has little impact on games as you can clearly see how close the i5 8600k and the i7 8700k are.
https://www.techpowerup.com/reviews/Intel/Core_i9_9900K/12.html
Holidays are over.. you are a bit late to the party.
Last edited:
DuronBurgerMan
[H]ard|Gawd
- Joined
- Mar 13, 2017
- Messages
- 1,340
Hardware.fr did a good comparison and average across all games. The 2700X was faster than the 8400 by a pubic hair. The 2600X was slower than the 8400 by the very same pubic hair. Of course, this is irrelevant to IPC - the 2700X and 2600X clocked higher than the 8400. But do get it right. The 7600k does get murdered though. 4 threads are not enough.