Why does Ryzen 7 1800X performs so poorly in games?

The results are wrong on Win7 with Ryzen, while Win10 is using all logical cores correctly - context of SMT enabled.
Other review sites show that even Win8 (not too different) will use all logical cores if SMT is not disabled.
Here is F1 2014 at DSO with the 6-core 4930K

F1-2014-CPU-Graph.jpg


Going by the vids the SMT related cores should be at 0% but they are not.
If I get time I will try to find other examples from different sites but it is late here.
Cheers
Just ran on my 8350 and it sometimes at points are equal across all cores but most of the time only one core in each module is higher than the other, like 47% on one 18% on the other. Took a screenshot but gotta find the little bugger. On the positive left it at normal 1080p and got 300fps avg, lol.
 
Before your argument comes valid you must consider core count. You mention price and performance but not getting 4 more cores for the same price or slightly higher.

If those cores aren't being used they aren't worth anything. Besides, my argument clearly already takes the extra cores into account considering I specifically said any productivity work that makes use of more cores will almost certainly be better suited to the 1700.

If a chip is 10-20% slower in a task and the same price, it doesn't matter that it had more cores. It still isn't performing as well in that particular task.

If you are primarily gaming and in the $300+ tier, go Intel. If you are primarily doing heavily multi threaded productivity work, go AMD for anything short of the absolute top performance tiers.

How is this so hard to understand? Why do some people feel the need to start spouting nonsense like "so long as you get 60fps in current games anything else is irrelevant". Completely ignoring minimums, high fps monitors and future performance.
 
How is this so hard to understand? Why do some people feel the need to start spouting nonsense like "so long as you get 60fps in current games anything else is irrelevant". Completely ignoring minimums, high fps monitors and future performance.

Consumers need to feel they didn't make a bad choice in their purchase and often make up rediculous notions to justify their spending in an attempt to avoid buyers remorse.
 
If you are primarily gaming and in the $300+ tier, go Intel. If you are primarily doing heavily multi threaded productivity work, go AMD for anything short of the absolute top performance tiers.
Yep, just how it is.

As an aside, when was the last time I heard someone say "go AMD" for anything at the high end. Even productivity.

I'm more used to hearing "yeah I bought that 8320 because it was super cheap and I just play porn".
 
I wonder how many porn videos you can open on 4 vs 8 cores before it starts lagging.
 
If those cores aren't being used they aren't worth anything. Besides, my argument clearly already takes the extra cores into account considering I specifically said any productivity work that makes use of more cores will almost certainly be better suited to the 1700.

If a chip is 10-20% slower in a task and the same price, it doesn't matter that it had more cores. It still isn't performing as well in that particular task.

If you are primarily gaming and in the $300+ tier, go Intel. If you are primarily doing heavily multi threaded productivity work, go AMD for anything short of the absolute top performance tiers.

How is this so hard to understand? Why do some people feel the need to start spouting nonsense like "so long as you get 60fps in current games anything else is irrelevant". Completely ignoring minimums, high fps monitors and future performance.
No it is still short sighted saying just go 4 cores. It isn't a bad process just not indicative of the whole. When comparing 2 chips at identical pricing then you add in the positives and negatives of each. In this case the 1700 is within spitting distance most of the time for games but out performs that same 4 core in everything else. And as far as minimums in most real world use cases the 1700 wins out, it just trails with the avg frame. Even in a lot of the frame time graphs the 1700 lead the 7700k. Add to that with games using more cores and your desire for high frame rates that same 7700k would have an even harder time with running anything else at the same time. Hell I fit that definition of "just a gamer" but I have ripped dvds and converted video. And the best part when I didi was I was still able to game and watch Netflix at the same time, all this on a lowly 8350.

So in regards to your original post, you didn't give hardly any weight to having more cores. Given both at the same price, both OCed to max the 7700k only has the gaming numbers in high refresh rate scenarios, the 1700 wins out nearly every where else. Besides I, as well as others, like to have it and not need it (8 cores), than need it and not have it.
 
But, and this is the baffling part, Win 7 shows in most benches with pics that it doesn't populate the SMT cores almost at all. Now in windows 10 it shows populating all cores + SMT equally.

Analysis performed by reviews sites has demonstrated that manually disabling SMT improves performance by 1% (minimum fps) and 3% (average fps) and AMD has stated officially that there is no issue with SMT scheduling.
 
Consumers need to feel they didn't make a bad choice in their purchase and often make up rediculous notions to justify their spending in an attempt to avoid buyers remorse.

Indeed! It is interesting seeing certain people in a perennial denial-mode, and not only rejecting all reviews analysis or measurements that disprove their viewpoints but even rejecting official statements from AMD!
 
Last edited:
Indeed! It is interesting seeing certain people in a perennial denial-mode, and not only rejecting all reviews analysis or measurements that disprove their viewpoints but even rejecting official statements from AMD itself that contradict their viewpoint.

AMD is so great they are actually holding back! *Inserts site with amazing scores and a joker clip on youtube* The truth can not be held back any longer! /s


BTW I think that Ryzen is a hit for AMD, just not the hit some are making it to be.
 
Just ran on my 8350 and it sometimes at points are equal across all cores but most of the time only one core in each module is higher than the other, like 47% on one 18% on the other. Took a screenshot but gotta find the little bugger. On the positive left it at normal 1080p and got 300fps avg, lol.
It does not necessarily equalise the load and depends upon the application, games are pretty dreadful for equalising load across all logical cores because they are just designed with some threads having much more importance.
But you show your system is using all cores because in his example it is at 0%.
TBH the guy should not had used games to test all cores in use and show it not being same as Win 10, why he did not use Cinebench or something similar in the vids to make a point I do not know.
Call of Duty Advanced Warfare seems to be more balanced across logical cores than some other games, compare it to my earlier example of F1 2014 and it is better, this is on Win8 at DSO: http://www.dsogaming.com/pc-perform...uty-advanced-warfare-pc-performance-analysis/

Call-of-Duty-Advanced-Warfare-CPU-graph.jpg


Cheers
 
Last edited:
Analysis performed by reviews sites has demonstrated that manually disabling SMT improves performance by 1% (minimum fps) and 3% (average fps) and AMD has stated officially that there is no issue with SMT scheduling.
Had you actually read the context here, this isn't about any performance increase more so than allocation nuances and whether the schedulers are working as they should.

As an added question... do you believe everything AMD states?
 
It does not necessarily equalise the load and depends upon the application, games are pretty dreadful for equalising load across all logical cores because they are just designed with some threads having much more importance.
But you show your system is using all cores because in his example it is at 0%.
TBH the guy should not had used games to test all cores in use and show it not being same as Win 10, why he did not use Cinebench or something similar in the vids to make a point I do not know.
Call of Duty Advanced Warfare seems to be more balanced across logical cores than some other games, compare it to my earlier example of F1 2014 and it is better, this is on Win8 at DSO: http://www.dsogaming.com/pc-perform...uty-advanced-warfare-pc-performance-analysis/

Call-of-Duty-Advanced-Warfare-CPU-graph.jpg


Cheers
I guess my point and question was in the stark difference between 7 and 10 in those few benches. Again no big deal but does speak to whether the schedule is working as intended or no. Could be 7 is wrong being the reports these CPUs wouldn't have 7 support from AMD.
 
I guess my point and question was in the stark difference between 7 and 10 in those few benches. Again no big deal but does speak to whether the schedule is working as intended or no. Could be 7 is wrong being the reports these CPUs wouldn't have 7 support from AMD.
But you proved the video is wrong in how it presents its context/conclusion (if he really feels Win7 is working fine and Win10 is a problem) as you have all logical cores in use even if partially :)
If your alternate logical cores are not 0% they are being used, the problem is you need a game that scales and uses many threads to really test it, or better a mult-threaded application/benchmark that is not a game (academic though because we have shown SMT is working).
And then DSO proves the vid wrong as again the games use all logical cores in Win8 showing there is no problems with Win10 SMT, let alone PCPer also proving this with their own basic apps to specifically look at this.
Ryzen is not fully supported by Win7, and has limitations from the conclusion from Looncraz who has done quite extensive testing comparing Win 7 to Win 10 at Anandtech forums.

What is interesting is how performance increases on Win7 compared to Win10 while using Nvidia GPU.
Cheers
 
Last edited:
Had you actually read the context here, this isn't about any performance increase more so than allocation nuances and whether the schedulers are working as they should.

As an added question... do you believe everything AMD states?
You don't get it both ways. You can't insist AMD was being honest before but lying about the scheduler not being the issue, but I digress. The fact is AMD does not strike me as a company with any problems laying the blame on Microsoft not to mention virtually all tests show the scheduler is working fine. The fact this only effects games lends credence to the idea it is game specific optimizations that will be needed not some magical scheduler update that AMD themselves has admitted works correctly.
 
Analysis performed by reviews sites has demonstrated that manually disabling SMT improves performance by 1% (minimum fps) and 3% (average fps)

That's a little misleading. Those percentages are the overall effect across every game they tested, which isn't really the interesting part. In the majority of games, disabling SMT either has a negative or no effect, which is the expected behavior. There are, however, a couple games where performance improves around 15%, which is totally odd.

Don't get me wrong, I don't subscribe to the idea there's some magic bullet lingering out there. But there are a couple titles out there where SMT seems to be bugged, probably due to the game's coding itself.
 
Don't get me wrong, I don't subscribe to the idea there's some magic bullet lingering out there. But there are a couple titles out there where SMT seems to be bugged, probably due to the game's coding itself.

Are those same games also HT bugged?
 
If you are primarily gaming and in the $300+ tier, go Intel. If you are primarily doing heavily multi threaded productivity work, go AMD for anything short of the absolute top performance tiers.

The real kicker is when you're like me... and you do both those things. There's a thread of mine in here where I'm talking through which one to buy. Though I skew more towards working, these days, and so ultimately I chose Ryzen. But it wasn't an easy choice. The 7700k is a very good CPU too.

You just gotta be honest with yourself, step away from the fact that we all kinda hate on Intel, because they play their stupid pricing games and throw their weight around like a monopoly. Yeah. That sucks. And they earned some well-justified hate with that behavior. But that doesn't mean you go off and buy an AMD chip if it's wrong for your use case. So here's the BurgerMan no bullsh*t decision tree for people who can afford to be spendy on their CPUs.

1. Are you made of cash and do you want the fastest sh*t you can buy for pure bragging rights? If yes, buy 6950k and go home. If no, continue.
2. Is this box primarily for gaming? If yes, buy 7700k. If no, continue.
3. Is this box primarily for productivity and content creation? If yes, buy Ryzen 8 core flavor of choice. If no, continue.
4. Is this box a mixed bag of gaming and content creation? You could choose either 7700k or Ryzen 8 core of choice, depending on which you do more.

Done deal. I'm sure if you're on a sub-$300 CPU budget, the calculus is different. But there's the matrix to follow. For my case, it was option 4, and I leaned toward work more than games, so Ryzen it was.

Kudos to AMD, though, for making it back on the list in ANY form. Bulldozer would have made it on the "would you like a 220 watt steaming pile of sh*t" list. Ryzen has a solid use case as a high performance part. Good times.
 
You don't get it both ways. You can't insist AMD was being honest before but lying about the scheduler not being the issue, but I digress. The fact is AMD does not strike me as a company with any problems laying the blame on Microsoft not to mention virtually all tests show the scheduler is working fine. The fact this only effects games lends credence to the idea it is game specific optimizations that will be needed not some magical scheduler update that AMD themselves has admitted works correctly.
Wrong. Plain and simple. You assume I said there was performance here and there one way or the other, and maybe in a small way yes. But if the scheduler isn't as efficient as it could be then there is the issue of instantaneous performance increases.

For instance core parking wont kill those AVG fps, generally does lower minimums a point or more. Based on that some may say it doesn't impact performance. But I and numerous others, AMD users and Intel alike, will tell you it makes a huge impact on gameplay as far as smoothness and that is a performance issue albeit not generally seen in bench results to a degree higher than common variance.

And you are doing the same thing as he is, so I will ask you the same question: Do you believe everything AMD says?
 
That's a little misleading. Those percentages are the overall effect across every game they tested, which isn't really the interesting part. In the majority of games, disabling SMT either has a negative or no effect, which is the expected behavior. There are, however, a couple games where performance improves around 15%, which is totally odd.

Don't get me wrong, I don't subscribe to the idea there's some magic bullet lingering out there. But there are a couple titles out there where SMT seems to be bugged, probably due to the game's coding itself.

It is not misleading, it is just the conclusion of the review:

From the 16 games tested we see that disabling SMT on the 1800X resulted in 3% more performance for the average frame rate and just 1% for the minimum.

Note that analysis was made when the forums were still filled with the idea that a magic SMT patch was going to solve Ryzen gaming problems. If you check the motivation, the reviewer wanted to test a claim made on Anandtech forums.

Of course, disabling SMT improves performance in some titles, reduces performance in other titles and doesn't change anything in the rest. That is why overall, SMT only provides 1% or 3% variations. If you are only interested in some specific title, then you have to check if that title benefits or not from disabling SMT:

Ryzen's overall average performance didn't look great but that's because the results in some games were actually hurt by turning off SMT. Those titles included Arma 3, Battlefield 1, Mafia III and Watch Dogs 2.

Conversely, disabling SMT boosted performance to some extent in Deus Ex: Mankind Divided, F1 2016, Far Cry Primal, Gears of War 4, Grand Theft Auto V, Overwatch and Total War: Warhammer, while we saw virtually no performance change in Civilization IV, For Honor, Hitman, Mirror's Edge Catalyst and The Division.
 
Wrong. Plain and simple. You assume I said there was performance here and there one way or the other, and maybe in a small way yes. But if the scheduler isn't as efficient as it could be then there is the issue of instantaneous performance increases.

For instance core parking wont kill those AVG fps, generally does lower minimums a point or more. Based on that some may say it doesn't impact performance. But I and numerous others, AMD users and Intel alike, will tell you it makes a huge impact on gameplay as far as smoothness and that is a performance issue albeit not generally seen in bench results to a degree higher than common variance.

And you are doing the same thing as he is, so I will ask you the same question: Do you believe everything AMD says?

Reviews have demonstrated the scheduler works fine and is not crippling Ryzen performance in any way. AMD has made an official statement admitting that their own tests prove that the scheduler works fine. RyZen problems are microarchitectural.

How many more time denying reality?
 
Reviews have demonstrated the scheduler works fine and is not crippling Ryzen performance in any way. AMD has made an official statement admitting that their own tests prove that the scheduler works fine. RyZen problems are microarchitectural.

How many more time denying reality?
How many times you going to ignore the questions I ask of you...

1 What do you have to say to the owners that thus far have reported no performance issues in the games they play?

2 Do you believe everything AMD states?

Try answering those or is it by answering them you give away your innate bias and true intent.
 
And you are doing the same thing as he is, so I will ask you the same question: Do you believe everything AMD says?
That is a completely irrelevant question to the issue at hand and a pretty convenient red herring. So I ask the question why does it matter if I believe everything or don't believe everything AMD says?
 
What am told that new revisions of Ryzen mainstream parts do not exhibit the same issues with gaming as first version Chips, probably why AMD didn't really contest microsofts schedular as the issue was more the chip. The problem is if you have already bought you are locked in to that performance as no firmware is able to change the performance to the later revisions.

Also X399 is confirmed as a new platform for AMD, I am told it is a LGA platform with up to quad channel support, the top end SKU is a 16 Core / 32 Thread processor for ~1000USD but has 3 confirmed lesser SKU's which AMD will term its HEDT platform to rival Intels X99 and X299 platform. Chiphell leaked the information and after some talks with people in places it was approved but the specs on chiphell were incorrect, mostly the 8 channel quoted is actually 8 DIMM slots for quad channel.
 
What am told that new revisions of Ryzen mainstream parts do not exhibit the same issues with gaming as first version Chips, probably why AMD didn't really contest microsofts schedular as the issue was more the chip. The problem is if you have already bought you are locked in to that performance as no firmware is able to change the performance to the later revisions.

Also X399 is confirmed as a new platform for AMD, I am told it is a LGA platform with up to quad channel support, the top end SKU is a 16 Core / 32 Thread processor for ~1000USD but has 3 confirmed lesser SKU's which AMD will term its HEDT platform to rival Intels X99 and X299 platform. Chiphell leaked the information and after some talks with people in places it was approved but the specs on chiphell were incorrect, mostly the 8 channel quoted is actually 8 DIMM slots for quad channel.
Seems a bit far fetched, what are your sources?
 
Hopefully they give the mobo makers much more lead time this time around.

The mobo makers had lots of time, but not lots of information or support from AMD and frankly they weren't all that into this launch themselves. Things have changed.
 
The mobo makers had lots of time, but not lots of information or support from AMD and frankly they weren't all that into this launch themselves. Things have changed.

Yeah he said they were extremely unprepared for first revision, but no sooner did that finish AMD bombarded them with revision 2 and 3 silicon. All i know is that the issues of first stepping have been remedied, but sadly for early adopters, you are locked in.

As for x399 it was confirmed to exist with the 16 core part. The leak by TR on 3400+ speed support was actually for this platform. AMD hid it well and this makes the R7 and X370 basically mid level
 
What am told that new revisions of Ryzen mainstream parts do not exhibit the same issues with gaming as first version Chips, probably why AMD didn't really contest microsofts schedular as the issue was more the chip. The problem is if you have already bought you are locked in to that performance as no firmware is able to change the performance to the later revisions.

Also X399 is confirmed as a new platform for AMD, I am told it is a LGA platform with up to quad channel support, the top end SKU is a 16 Core / 32 Thread processor for ~1000USD but has 3 confirmed lesser SKU's which AMD will term its HEDT platform to rival Intels X99 and X299 platform. Chiphell leaked the information and after some talks with people in places it was approved but the specs on chiphell were incorrect, mostly the 8 channel quoted is actually 8 DIMM slots for quad channel.

They already have a new revision working? When is it available? I'm glad I held out on my purchase. I have all my new system now but the CPU and mainboard.
 
What am told that new revisions of Ryzen mainstream parts do not exhibit the same issues with gaming as first version Chips, probably why AMD didn't really contest microsofts schedular as the issue was more the chip. The problem is if you have already bought you are locked in to that performance as no firmware is able to change the performance to the later revisions.

Also X399 is confirmed as a new platform for AMD, I am told it is a LGA platform with up to quad channel support, the top end SKU is a 16 Core / 32 Thread processor for ~1000USD but has 3 confirmed lesser SKU's which AMD will term its HEDT platform to rival Intels X99 and X299 platform. Chiphell leaked the information and after some talks with people in places it was approved but the specs on chiphell were incorrect, mostly the 8 channel quoted is actually 8 DIMM slots for quad channel.

So do they mean they are quietly putting new r7's into the channel that have these issues fixed under the same naming scheme (1700,1700x,1800x) or are these going to be called something else? I just find it hard to believe they already have a 2nd revision chip ready to go so soon after launch. Anybody who pre-ordered or bought in the first few months would be pissed beyond belief. I cant imagine they could do this without offering an exchange or recall of some sort.
 
I cant imagine they could do this without offering an exchange or recall of some sort.

1725X coming your way in no time... I don't believe this though, chip revs take a LOT of time. I do believe in another platform to handle Naples and its spin offs.
 
So do they mean they are quietly putting new r7's into the channel that have these issues fixed under the same naming scheme (1700,1700x,1800x) or are these going to be called something else? I just find it hard to believe they already have a 2nd revision chip ready to go so soon after launch. Anybody who pre-ordered or bought in the first few months would be pissed beyond belief. I cant imagine they could do this without offering an exchange or recall of some sort.

It sucks for sure but intel and nvidia will do the same, chips have to be sold and early adopters have to realize that they may be test subjects.

I have no idea what Branding will be used, all i know is that no firmware can fix it to later silicon. I think though Q2-3 most likely
 
1725X coming your way in no time... I don't believe this though, chip revs take a LOT of time. I do believe in another platform to handle Naples and its spin offs.
Naples is its own platform, X399 is also its own, as above LGA
 
So do they mean they are quietly putting new r7's into the channel that have these issues fixed under the same naming scheme (1700,1700x,1800x) or are these going to be called something else? I just find it hard to believe they already have a 2nd revision chip ready to go so soon after launch. Anybody who pre-ordered or bought in the first few months would be pissed beyond belief. I cant imagine they could do this without offering an exchange or recall of some sort.

Yeah i can imagine, I don't agree with it either and i hope AMD gives rebates
 
Yeah he said they were extremely unprepared for first revision, but no sooner did that finish AMD bombarded them with revision 2 and 3 silicon. All i know is that the issues of first stepping have been remedied, but sadly for early adopters, you are locked in.

As for x399 it was confirmed to exist with the 16 core part. The leak by TR on 3400+ speed support was actually for this platform. AMD hid it well and this makes the R7 and X370 basically mid level
Seems a bit odd to market the 1800x as a x99 killer to then release a more expensive x99 type variant.
 
Back
Top