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

We test at 3ghz, is that unfair? Clock speed is the same

Gaming at 3GHz

getgraphimg.php


About 20% behind Broadwell on average.
 
The R5 uses worse silicon and cannot clock so high as the R7 1800X. With half the number of cores/threads and lower clocks one would expect the R5 to game worse than the 1800X.

It is not causal that AMD did launch the R7 models first to generate a halo effect and try to hide the clock problems with quad-core chips. In fact this possibility was mentioned before launch:



https://www.cpchardware.com/intel-prepare-la-riposte-a-ryzen/
the 6/12 chip has been officially confirmed as 3.6-40ghz, so considering virtually no games take advantage off all 8 cores if they do fix some othe other problems before it launches it won't be a bad chip for a sub $300 price
 
Why? What mystery are you solving by underclocking chips?

Lower frequency puts more strain on processing performance, Floating point is put under stress. Also the 6950, 5960 and 1700 come out the box at 3 ghz so we can see what 30% clockspeed does to IPC on singlethread, but not much for parallelism
 
Lower frequency puts more strain on processing performance, Floating point is put under stress. Also the 6950, 5960 and 1700 come out the box at 3 ghz so we can see what 30% clockspeed does to IPC on singlethread, but not much for parallelism


err not exactly we will need to see the load of the cpu cores to see what is going on, to me with the IPC drop on a per core level because of the frequency drop should increase load on all cores across the board, but that doesn't seem to be what is happening, so in affect you are just knee capping Intel chips with AMD's not as much.
 
Gaming at 3GHz

getgraphimg.php


About 20% behind Broadwell on average.

That must be a percentage graph, most are inadequate, inaccurate and thumb sucks.

We roughly know where ryzen is, we know it cannot match skykabylake ipc but across all tests it is incredibly balanced. It is what my 5960X is to me
 
With the announcement about Naples it seems clear what AMD did here. AMD focused on building a server chip first and a desktop chip second. That's what they did with Bulldozer but this time it seems AMD did it right. Ryzen is to the Bulldozer as Core 2 was to the Pentium IV. AMD probably figured that games are mostly GPU limited and the actual gaming performance of Ryzen should be satisfactory. I think they probably made the correct call. The information we have about Naples suggests that this is a monster for the server market and that's where the money is.

I'm hoping AMD comes back at the enthusiast market with something to really challenge the HEDT market with something based on Naples. This wouldn't be unheard of as that's largely what AMD did with it's Athlon FX-51.
 
err not exactly we will need to see the load of the cpu cores to see what is going on, to me with the IPC drop on a per core level because of the frequency drop should increase load on all cores across the board, but that doesn't seem to be what is happening, so in affect you are just knee capping Intel chips with AMD's not as much.

The generational IPC gap Kaby has over older CPU's and ryzen remains consistent at all clocks so that is not true, it is not kneecaping in any way, and is great to see what clock for clock each arch has
 
With the announcement about Naples it seems clear what AMD did here. AMD focused on building a server chip first and a desktop chip second. That's what they did with Bulldozer but this time it seems AMD did it right. Ryzen is to the Bulldozer as Core 2 was to the Pentium IV. AMD probably figured that games are mostly GPU limited and the actual gaming performance of Ryzen should be satisfactory. I think they probably made the correct call. The information we have about Naples suggests that this is a monster for the server market and that's where the money is.

I'm hoping AMD comes back at the enthusiast market with something to really challenge the HEDT market with something based on Naples. This wouldn't be unheard of as that's largely what AMD did with it's Athlon FX-51.

I have built 7 ryzen rigs on different boards including a not so good ITX board, from ASUS to Biostar and these CPU's are good old fashioned American muscle while Kaby is more euro fashion car. Nothing like a rolling up supercharger.

I can make my music and compile it in no time on a 1700. Very impressive and i can play Resident Evil 7, if you guys haven't played it, you should
 
I have built 7 ryzen rigs on different boards including a not so good ITX board, from ASUS to Biostar and these CPU's are good old fashioned American muscle while Kaby is more euro fashion car. Nothing like a rolling up supercharger.

I can make my music and compile it in no time on a 1700. Very impressive and i can play Resident Evil 7, if you guys haven't played it, you should


Which is your preferred mobo thus far?
 
Lower frequency puts more strain on processing performance, Floating point is put under stress. Also the 6950, 5960 and 1700 come out the box at 3 ghz so we can see what 30% clockspeed does to IPC on singlethread, but not much for parallelism

err not exactly we will need to see the load of the cpu cores to see what is going on, to me with the IPC drop on a per core level because of the frequency drop should increase load on all cores across the board, but that doesn't seem to be what is happening, so in affect you are just knee capping Intel chips with AMD's not as much.

The generational IPC gap Kaby has over older CPU's and ryzen remains consistent at all clocks so that is not true, it is not kneecaping in any way, and is great to see what clock for clock each arch has

All of that is academic and at the end of the day IPC is not the only or most important measurement of performance. The total package has to be considered and that means clock speeds. Lowering all the clock speeds might be a nice test for IPC but in no ways is representive of performance of either chip.
 
With the announcement about Naples it seems clear what AMD did here. AMD focused on building a server chip first and a desktop chip second. That's what they did with Bulldozer but this time it seems AMD did it right. Ryzen is to the Bulldozer as Core 2 was to the Pentium IV. AMD probably figured that games are mostly GPU limited and the actual gaming performance of Ryzen should be satisfactory. I think they probably made the correct call. The information we have about Naples suggests that this is a monster for the server market and that's where the money is.

I'm hoping AMD comes back at the enthusiast market with something to really challenge the HEDT market with something based on Naples. This wouldn't be unheard of as that's largely what AMD did with it's Athlon FX-51.

I have built 7 ryzen rigs on different boards including a not so good ITX board, from ASUS to Biostar and these CPU's are good old fashioned American muscle while Kaby is more euro fashion car. Nothing like a rolling up supercharger.

I can make my music and compile it in no time on a 1700. Very impressive and i can play Resident Evil 7, if you guys haven't played
Which is your preferred mobo thus far?

I am extremely msi so the carbon gamer is nice, best out the box is the gigabyte aorus but surprisingly tje asrock was good value.
 
With the announcement about Naples it seems clear what AMD did here. AMD focused on building a server chip first and a desktop chip second. That's what they did with Bulldozer but this time it seems AMD did it right. Ryzen is to the Bulldozer as Core 2 was to the Pentium IV. AMD probably figured that games are mostly GPU limited and the actual gaming performance of Ryzen should be satisfactory. I think they probably made the correct call. The information we have about Naples suggests that this is a monster for the server market and that's where the money is.

I'm hoping AMD comes back at the enthusiast market with something to really challenge the HEDT market with something based on Naples. This wouldn't be unheard of as that's largely what AMD did with it's Athlon FX-51.


Well the presentations AMD has done with Naples 32 core parts they putting them against Intel 22 (x2 so 44 )core last gen parts, so I wouldn't be so sure its a monster yet lol. Intel already has 24 core individual Xeons, and they are upping them to 28 core with Skylake E Xeons.

If I'm reading their marketing slides correctly, the numbers aren't adding up either

https://videocardz.com/67086/amd-previews-naples-server-processor-coming-q2-2017
 
Last edited:
Fair enough. We knew that Ryzen was going head to head with Kaby Lake but the real problem is that AMD is still playing catch up. What AMD had to face on Ryzen's launch day is one thing, what's coming next from Intel is another. Intel still has a big start on future architectures which Ryzen will have to face off against. AMD already falls behind Skylake and Kaby Lake in terms of IPC and clock speeds. Ryzen does fairly well against Broadwell-E in some applications but as you said, Skylake-E is coming. One thing Naples definitely has going for it is the 64 PCIe lanes. That's fucking fantastic.
 
Well the presentations AMD has done with Naples 32 core parts they putting them against Intel 8 (x2 so 16 )core last gen parts, so I wouldn't be so sure its a monster yet lol. Intel already has 24 core individual Xeons, and they are upping them to 28 core with Skylake E Xeons.

If I'm reading their marketing slides correctly, the numbers aren't adding up either

https://videocardz.com/67086/amd-previews-naples-server-processor-coming-q2-2017
Cost and availability will be the big factor here. If the price is right and power consumption is at or slightly better than the Xeons they may see success here, but they also need to be able to pump them out in the volumes necessary to fulfil orders for server farms.
 
Fair enough. We knew that Ryzen was going head to head with Kaby Lake but the real problem is that AMD is still playing catch up. What AMD had to face on Ryzen's launch day is one thing, what's coming next from Intel is another. Intel still has a big start on future architectures which Ryzen will have to face off against. AMD already falls behind Skylake and Kaby Lake in terms of IPC and clock speeds. Ryzen does fairly well against Broadwell-E in some applications but as you said, Skylake-E is coming. One thing Naples definitely has going for it is the 64 PCIe lanes. That's fucking fantastic.


Yeah it has the features that Intel won't have till after Purely (Purely is still stuck at 48 PCI-e lanes too)

Actually no its not stuck at 48, its because of Skylake - E its stuck at 48?
 
Cost and availability will be the big factor here. If the price is right and power consumption is at or slightly better than the Xeons they may see success here, but they also need to be able to pump them out in the volumes necessary to fulfil orders for server farms.


Only if their performance can out match Intel, just ram costs alone are more expensive than the CPU costs, so they will need to show that switching over, in the long run will save the company money than the short term.
 
Well the presentations AMD has done with Naples 32 core parts they putting them against Intel 22 (x2 so 44 )core last gen parts, so I wouldn't be so sure its a monster yet lol. Intel already has 24 core individual Xeons, and they are upping them to 28 core with Skylake E Xeons.

If I'm reading their marketing slides correctly, the numbers aren't adding up either

https://videocardz.com/67086/amd-previews-naples-server-processor-coming-q2-2017

http://www.anandtech.com/Gallery/Album/5506#23

Stupid Videocardz, omitting the annotations slide. This seems like the whole slideshow. TheTechReport has videos of demos taking place.
 
Cost and availability will be the big factor here. If the price is right and power consumption is at or slightly better than the Xeons they may see success here, but they also need to be able to pump them out in the volumes necessary to fulfil orders for server farms.

Without a doubt.
 
Fair enough. We knew that Ryzen was going head to head with Kaby Lake but the real problem is that AMD is still playing catch up. What AMD had to face on Ryzen's launch day is one thing, what's coming next from Intel is another. Intel still has a big start on future architectures which Ryzen will have to face off against. AMD already falls behind Skylake and Kaby Lake in terms of IPC and clock speeds. Ryzen does fairly well against Broadwell-E in some applications but as you said, Skylake-E is coming. One thing Naples definitely has going for it is the 64 PCIe lanes. That's fucking fantastic.

It is bad to necessarily assume Intel will deliver, they haven't advanced on anything but clocks since haswell and there is a limit to how high you can push clockspeed. Coffee Lake is 14nm again and from what i have poked around in, there are no 6 cores for mainstream, iGPU is to big. But the surprise may be that the iGPU bares a Vega branding.

Skylake X marginally better than Broadwell E with slightly higher clocks. The issue is the core architecture is what 11 years old and well bedded but IPC is stagnant.

From AMD side its hush but a node change is highly likely to a more stable power node.
 
It is bad to necessarily assume Intel will deliver, they haven't advanced on anything but clocks since haswell and there is a limit to how high you can push clockspeed. Coffee Lake is 14nm again and from what i have poked around in, there are no 6 cores for mainstream, iGPU is to big. But the surprise may be that the iGPU bares a Vega branding.

Skylake X marginally better than Broadwell E with slightly higher clocks. The issue is the core architecture is what 11 years old and well bedded but IPC is stagnant.

From AMD side its hush but a node change is highly likely to a more stable power node.


its more of a platform issue that the CPU's, Skylake E will not be able to remedy this because it wasn't made to take advantage of all of Purely's capabilities (at least that what it looks like from the tidbits of info that have spilled out), but Cannon Lake next year will.
 
It is bad to necessarily assume Intel will deliver, they haven't advanced on anything but clocks since haswell and there is a limit to how high you can push clockspeed. Coffee Lake is 14nm again and from what i have poked around in, there are no 6 cores for mainstream, iGPU is to big. But the surprise may be that the iGPU bares a Vega branding.

Skylake X marginally better than Broadwell E with slightly higher clocks. The issue is the core architecture is what 11 years old and well bedded but IPC is stagnant.

From AMD side its hush but a node change is highly likely to a more stable power node.

My point is that Skylake-E would have a moderate enough IPC lead to swing most if not all of those performance benchmarks Ryzen is actually good at against the 6900K back in Intel's favor. Given the overclocking performance of Skylake and Kaby Lake, its possible we will see parts that clock higher. We've seen that a TIM change alone can produce higher clock speeds on CPUs that otherwise end up being middle of the road overclockers. Intel may be able to make that change in a revision or even in the production versions of its next CPU SKUs. Intel still has a massive clock speed and a modest IPC advantage over AMD.

I think Intel got caught with their pants down somewhat but this isn't another Athlon vs. Pentium IV scenario.
 
its more of a platform issue that the CPU's, Skylake E will not be able to remedy this because it wasn't made to take advantage of all of Purely's capabilities (at least that what it looks like from the tidbits of info that have spilled out), but Cannon Lake next year will.

CPU performance aside, this is actually where Naples is seriously looking good.
 
Gaming at 3GHz

getgraphimg.php


About 20% behind Broadwell on average.
We all know it's slower. But common man. We are talking 160+ here. Jesus. Get your point. But it just proves it's a well balanced chip. We expected it to slower across the board 20%. I'll take 20% slower in games at low resolution when it still gives me 100+ Vs 120 fps, if you know what I mean. It's still one hell of a competitive architecture given how much R&D budget intel has over amd.
 
We all know it's slower. But common man. We are talking 160+ here. Jesus. Get your point. But it just proves it's a well balanced chip. We expected it to slower across the board 20%. I'll take 20% slower in games at low resolution when it still gives me 100+ Vs 120 fps, if you know what I mean. It's still one hell of a competitive architecture given how much R&D budget intel has over amd.

The best thing is not to engage him, for your sanity
 
the 6/12 chip has been officially confirmed as 3.6-40ghz, so considering virtually no games take advantage off all 8 cores if they do fix some othe other problems before it launches it won't be a bad chip for a sub $300 price

Both CPCHardware and me discussed quad-cores, because then certain people was pretending that AMD would release quad-cores with base clocks of 4.2GHz or higher. Also note posters who I was replying here mentioned quad-cores like the R5 1500X. I was replying that.

No doubt that 6-core RyZen at 3.6GHz will be a preferred option over the 1800X or the 1700X for gaming.
 
Obviously you need to go look at the 6900k on Newegg or any other site... it lists best gaming experience or something along those lines. It was posted in one of these threads you have been squatting in so I know you have seen it.

Newegg doesn't manufacture those chips, they are a reseller. C'mon man, use some fucking common sense.
 
My point is that Skylake-E would have a moderate enough IPC lead to swing most if not all of those performance benchmarks Ryzen is actually good at against the 6900K back in Intel's favor. Given the overclocking performance of Skylake and Kaby Lake, its possible we will see parts that clock higher. We've seen that a TIM change alone can produce higher clock speeds on CPUs that otherwise end up being middle of the road overclockers. Intel may be able to make that change in a revision or even in the production versions of its next CPU SKUs. Intel still has a massive clock speed and a modest IPC advantage over AMD.

I think Intel got caught with their pants down somewhat but this isn't another Athlon vs. Pentium IV scenario.
Actually it is closer to this than any other. The Pentium 4 had a performance advantage over the Athlon, sometimes AMD had to fall back a bit. I sure people here remember the Palomino days. MAD with their PR ratings and barely keeping up in a few tasks. But even in a power hungry time the P4 was hot, costly, and just not enough of a difference for a lot of people to pick it up.

This isn't like the Athlon vs. the PIII or the Athlon64 and certainly not the Athlon 64x2 (the best days of AMD).

People expecting any major changes in SL-X don't know Intel or the process. They might drop the prices slightly on their new CPU's, but don't expect much. They aren't going to make the platform any cheaper. These are server chips dressed as gamer chips by putting them on a "gamer" chipset.. Intel isn't going to trash their server chip margins to chase AMD on the desktop. The 6 core might hover between the R5 and R7. Maybe the 8 core drops to $800. The higher ones might come down but wouldn't count on it. You are still going to spend $1k to match the productivity. Now some of this changes with CL, but Intel isn't going to stop the presses, put get new silicon going for a 8C chip to battle AMD. The 6C desktop part is going to be priced at $350+ and will still be battling (though maybe winning more) against a 8c16t part. All the while setting up to deliver world beating mobile and SFF parts. Which will allow them to sell in the biggest market while waiting for the Zen 2k parts. This window looks small because of tunnel vision on single aspects of performance. Realistically this is smaller piece of the pie than most people recognize and this platform even without the clock and not being quite there in IPC, is more of platform introduction (with great value) that will open the door to design Wins in the markets that really matter ones that AMD due to this platform is set to be much better options than their alternatives (even if it doesn't get them as many wins as they should). By the time AMD should be worried about falling behind Intel in pure performance, they will be hitting their refresh cadence and they just need stay at or make up a little bit of ground on Intel per refresh.
 
The Athlon 64 destroyed the Pentium IV CPUs of the day. We are not in that situation.
 
The Athlon 64 destroyed the Pentium IV CPUs of the day. We are not in that situation.
You said Athlon which did run against the P4 for awhile. Thunderbird, Thuroughbred, Thuroughbred B, Palimino, and the Barton before the Athlon64.

The K7 competed against the Pentium 4 longer than the K8.
 
You said Athlon which did run against the P4 for awhile. Thunderbird, Thuroughbred, Thuroughbred B, Palimino, and the Barton before the Athlon64.

The K7 competed against the Pentium 4 longer than the K8.

Yes it did and that wasn't as bad an asswhipping as the Athlon 64 gave it.
 
We all know it's slower. But common man. We are talking 160+ here. Jesus. Get your point. But it just proves it's a well balanced chip. We expected it to slower across the board 20%. I'll take 20% slower in games at low resolution when it still gives me 100+ Vs 120 fps, if you know what I mean. It's still one hell of a competitive architecture given how much R&D budget intel has over amd.

Reviewers are performing those tests for two reasons: (i) to provide data to people that play games with those settings and (ii) to provide the so-called CPU tests, which help us to estimate future performance.

Note that the 20% was for both chips at same clocks. BDW-E overclocks better and the gap can increase to 100FPS vs 130FPS. You are right on that this difference is unnoticeable for most people, but a more demanding future title could reduce the absolute FPS and be noticeable.

And the gaming performance is better for Kabylake i7. Not strange that most (all?) reviewers are noticing that Kabylake i7 is a better choice for gaming.
 
Last edited:
Yes it did and that wasn't as bad an asswhipping as the Athlon 64 gave it.
Oh sure. I am just saying that while the K7 initially annoyed the hell out of Intel because AMD came in with a strong chip while Intel was finalizing their new arch. That this is closer to what the Athlon had to put up with most of it life. Really good in many spots, but always wavering between beating Intel and falling behind. The K8 on the other hand just plained owned it's entire life and it was what Intel was doing during this time frame that always make me sad each time I buy a CPU of theirs. I should have seen it in like 50% of the computers out there instead less than 10%.
 
Reviewers are performing those tests for two reasons: (i) to provide data to people that play games with those settings and (ii) to provide the so-called CPU tests, which help us to estimate future performance.

Note that the 20% was for both chips at same clocks. BDW-E overclocks better and the gap can increase to 100FPS vs 130FPS. You are right on that this difference is unnoticeable for most people, but a more demanding future title could the absolute FPS and be noticeable.

But that again isn't how it works. Ryzen isn't losing because the game puts a bigger demand on the CPU at this point. All it does it is cause the GPU to ask the CPU for page refreshes as fast as it can handle them and for the most part no processor in these tests are sitting at 99%. The fact is more CPU stressful games are asking the system for more actual work to get done. When that happens it's usually a developer actually leveraging MT tasks to increase CPU processing to handle things like Physics, AI, and other CPU used tasks. When those things happen (like BF MP) the Ryzen shines.

On top of all that. Outside uber competitive CSgo (pro) players, all most everyone keeps the GPU as the bottleneck (for looks). That shift keeps moving and as long as the CPU can send the workload needed for GPU, then there isn't an issue, and these again have for some reason I don't understand actually have Ryzen very very very slightly ahead (normal variance if it wasn't so wide spread). That 2 FPS at 1440 high probably matters more than 20 FPS at 1080p low.
 
Newegg doesn't manufacture those chips, they are a reseller. C'mon man, use some fucking common sense.
Don't feel like looking too hard or wasting time on it but this is what is on INTELs own site:

Equipped with Intel® Turbo Boost Technology 2.01 2, your computer will have the unprecedented power and responsiveness to help your productivity soar. Experience fantastic entertainment and gaming, seamless 4K Ultra HD, and 360 video– all with the lightning speed data transfers of Thunderbolt™ 3 technology.
 
Back
Top