Discussion AMD Ryzen MegaThread! FAQ and Resources

[goldstone77]

[video="https://www.youtube.com/watch?v=UfNMn7RWgLw"][/video]
AMD Ryzen 5 1600 vs. Intel Core i7-7800X: 30 Game Battle!
Hardware Unboxed
Published on Jul 21, 2017

Ryzen 1600 is tough to beat when it comes to price, performance, and power consumption.
Price has increase on this model to around $210, but B350 motherboards are still under $100.
https://pcpartpicker.com/product/mV98TW/amd-ryzen-5-1600-32ghz-6-core-processor-yd1600bbaebox

The 7800X is a tough sell when vs the 1600 when it comes to price, performance, and power consumption.
Price on the 7800X will cost north of $400, and a motherboard in the neighborhood of $250.
https://pcpartpicker.com/product/ymtWGX/intel-core-i7-7800x-35ghz-6-core-processor-bx80673i77800x

 

[juanrga]

Considering that they tested several GPU-bound games and some few games that were frame-capped, it is not surprising that overclocking the 7800X by 34% only brings 3% extra performance compared to stock 7800X. Also we know that SKL-X was run with stock clocks on the mesh interconnect, whereas the RyZen chip was tested with IF interconnect overclocked by using 3200MHz RAM. Overclocking the mesh on SKL-X can improve gaming performance by double digit percents compared to stock settings.

 

[TMTOWTSAC]

I think Hardware Unboxed was mostly talking about the gaming experience as it would be right now in existing games, admittedly at 1080p with a gtx 1080 ti. From that perspective, frame capping and GPU bound scenarios are real world factors gamers experience rather than artificial constraints. So it's a valid snapshot of the current gaming landscape. He didn't make any claims about future games or untapped performance, but I think that's what some people seem to be reading into it.

 

[jdwii]

Well from what i'm reading perhaps late this year we will really have great competition with updated Intel parts. I said it 100 times now 6 core skylake will compete very well to a 8 core ryzen. I see prices dropping and products getting better for everyone for the next few years.

 

[cdrkf]

@jdwii, I agree, I think thanks to ryzen 2018 will be the year hex core becomes the mainstream solution across all platforms. We're even seeing hex core laptop parts start to appear.

 

[jdwii]

https://postimg.org/image/rpqhlbz27/

Possibly leak benchmark of ryzen 3

Scores around a 4460

 

[goldstone77]

This board looks amazing, and they have more boards just follow the link.
ASRock and MSI X399 Motherboards For AMD Ryzen Threadripper CPUs Detailed – Pre-Orders Start in Less Than 24 Hours
By Hassan Mujtaba
4 hours ago
Gigabyte X399 AOURS Gaming 7 Motherboard
"The Gigabyte X399 AORUS Gaming 7 is the flagship motherboard of the lineup. In terms of design, the X399 AORUS Gaming 7 has a standard ATX design with the largest consumer socket we have seen to date, the TR4. The TR4 is a lot similar to the SP3 socket which is designed for the EPYC server platform and feature 4094 pins in total. The socket is surrounded by eight DDR4 DIMM slots that can support up to 128 GB of ram."

"The motherboard has a solid power delivery system which is delivered power through an 8 and 4 pin connector configuration. The power delivery system is cooled by two sets of aluminum based heatsinks. What’s interesting is that the X399 platform will feature up to 64 PCI-e Gen 3.0 lanes vs 44 max PCI-e Gen 3.0 lanes on Intel’s X299 platform. Expansion slots include four PCI-e 3.0 x16 slots and triple M.2 slots."

"For storage, we are looking at 8 SATA III ports along with tons of connectivity on the I/O panel and PCB for users. The rear panel includes 8 USB 3.0 ports, two USB 3.1 (Type A + Type C) ports, a 7.1 channel audio jack, a single LAN port and PS/2 connector. The motherboard is fully compliant with Gigabyte’s AURA RGB LED lighting system which looks tremendously great on their new HEDT flag ship products. If you pre-order the motherboard, you will be applicable to the Xpedition Pack promotion which grants you either a Kingston HyperX Cloud headset or a Steam $50 US voucher code which is splendid."
http://wccftech.com/amd-asrock-msi-gigabyte-x399-motherboard-ryzen-threadripper-cpus/

 

[goldstone77]

AMD Ryzen 5 1600 vs Intel Core i7-7800X: 30 Game Battle!
Page 9 : Power Consumption & The Verdict
By Steven Walton on July 21, 2017
Power Consumption & The Verdict

"I have to admit, this just seems wrong. I still can't get used to seeing AMD CPUs consuming less power than their Intel counterpart. It feels unnatural. Those power-hungry FX chips sat atop our power charts for so long and I'm glad we can finally move on from them.
For what seemed like a similar level of performance (we'll get to that in a moment), the R5 1600 consumed quite a lot less power compared to the 7800X. With both chips overclocked, the Ryzen system consumed 15% less power and 18% less before we did any overclocking -- a definitive win for AMD. Who predicted this at the beginning of the year? I'll admit it certainly wasn't me.
We tried to get to the point quickly on this one but there's no easy way to do that when benchmarking 30 PC games. Overall, we're impressed with the performance displayed by the Ryzen 5 1600. Granted, it was a good chip from day one but it seems like the few months of optimizations have made it even better.
Before we get to our great graph of averages, it's worth noting that after spending an extra two days confirming these results (mostly re-testing and comparing the Core i7-7700K and Ryzen 5 1600), we can say that while these chips are strangely close in performance, our figures are accurate.
Out of the box the R5 1600 is 13% slower than the more expensive 7700K. Keep in mind, there were a few GPU-limited games and even worse, a few that were frame-capped, they weren't limited to 60fps. Many had 120fps+ limits, so the Ryzen CPU was still pushing the GTX 1080 Ti hard."

"Once overclocked, the $215 R5 1600 was just 9% slower than the 7700K, while it matched the $415 7800X. I'm sure you're itching to see what the overall picture looks like so we won't wait any longer.
Just looking at the 7800X and R5 1600, here are the overclocked results when comparing the minimum frame rates. As we just saw, both averaged a minimum of 103fps across the 30 games tested. We can clearly see where the R5 1600 enjoyed some wins and suffered through a few loses."

"Some readers will undoubtedly declare that I'm biased towards a certain company, but if that was the case, there's no way you would see Deus Ex: Mankind Divided and Gears of War 4 in this list. Meanwhile, if I were biased towards Intel, I'd drop Total War: Warhammer and Civilization IV.
It's worth noting that the number of titles won and lost are equal, and even if we were to remove some of the worst performing games for either platform, it would only skew the results slightly toward either camp, which is the beauty of testing with such a massive sample of games.
For example, if we remove AMD's worst performing title, Gears of War 4, the R5 1600 would become just 1% faster than the 7800X. Remove AMD's second worst title, Deus Ex Mankind Divided using DX12, and the R5 1600 is still just 1% faster on average.
Please note that because there was a large difference in performance between running Deus Ex: Mankind Divided with DX11 and DX12 with the Ryzen CPU, I included both the results in this graph. However, I only included DX12 results for Total War: Warhammer and not the DX11 results since the R5 1600 was much faster in both tests. I favored DX12 as it's the newer API.
If you care at all about value, the Ryzen 5 1600 is clearly the way to go. This is why we recently named it the best value performance desktop CPU. It was unlikely that the Core i7-7800X was going to change that, but we hoped the performance would at least be a compelling reason to buy Intel's new six-core processor.
Ryzen will hit 4GHz with the box cooler but it will be a more mild experience with a $20 aftermarket cooler like the Cooler Master 212, so keep that in mind. The 7800X on the other hand cannot be overclocked to 4.7GHz using a 240mm AIO closed loop solution. Instead, it required a $380 custom loop setup to achieve that result."
https://www.techspot.com/review/1450-core-i7-vs-ryzen-5-hexa-core/page9.html

 

[randomizer]

Unless Coffee Lake is a big surprise I think a 1600 will be my next processor. I've been pondering a 1700 as well, but it's considerably more expensive and I don't think I will use the extra threads often. Then again, it might serve me better in 8 years.

 

[naturesninja]

Ryzen 3 on Userbenchmark:
http://cpu.userbenchmark.com/Compare/AMD-Ryzen-3-1300X-vs-AMD-Ryzen-3-1200/3930vs3931
http://cpu.userbenchmark.com/Compare/AMD-Ryzen-3-1300X-vs-Intel-Core-i5-7500/3930vs3648

 

[gamerk316]

AMD Ryzen 5 1600 vs Intel Core i7-7800X: 30 Game Battle!
Page 9 : Power Consumption & The Verdict
By Steven Walton on July 21, 2017
Power Consumption & The Verdict

"I have to admit, this just seems wrong. I still can't get used to seeing AMD CPUs consuming less power than their Intel counterpart. It feels unnatural. Those power-hungry FX chips sat atop our power charts for so long and I'm glad we can finally move on from them.
For what seemed like a similar level of performance (we'll get to that in a moment), the R5 1600 consumed quite a lot less power compared to the 7800X. With both chips overclocked, the Ryzen system consumed 15% less power and 18% less before we did any overclocking -- a definitive win for AMD. Who predicted this at the beginning of the year? I'll admit it certainly wasn't me.
We tried to get to the point quickly on this one but there's no easy way to do that when benchmarking 30 PC games. Overall, we're impressed with the performance displayed by the Ryzen 5 1600. Granted, it was a good chip from day one but it seems like the few months of optimizations have made it even better.
Before we get to our great graph of averages, it's worth noting that after spending an extra two days confirming these results (mostly re-testing and comparing the Core i7-7700K and Ryzen 5 1600), we can say that while these chips are strangely close in performance, our figures are accurate.
Out of the box the R5 1600 is 13% slower than the more expensive 7700K. Keep in mind, there were a few GPU-limited games and even worse, a few that were frame-capped, they weren't limited to 60fps. Many had 120fps+ limits, so the Ryzen CPU was still pushing the GTX 1080 Ti hard."

"Once overclocked, the $215 R5 1600 was just 9% slower than the 7700K, while it matched the $415 7800X. I'm sure you're itching to see what the overall picture looks like so we won't wait any longer.
Just looking at the 7800X and R5 1600, here are the overclocked results when comparing the minimum frame rates. As we just saw, both averaged a minimum of 103fps across the 30 games tested. We can clearly see where the R5 1600 enjoyed some wins and suffered through a few loses."

"Some readers will undoubtedly declare that I'm biased towards a certain company, but if that was the case, there's no way you would see Deus Ex: Mankind Divided and Gears of War 4 in this list. Meanwhile, if I were biased towards Intel, I'd drop Total War: Warhammer and Civilization IV.
It's worth noting that the number of titles won and lost are equal, and even if we were to remove some of the worst performing games for either platform, it would only skew the results slightly toward either camp, which is the beauty of testing with such a massive sample of games.
For example, if we remove AMD's worst performing title, Gears of War 4, the R5 1600 would become just 1% faster than the 7800X. Remove AMD's second worst title, Deus Ex Mankind Divided using DX12, and the R5 1600 is still just 1% faster on average.
Please note that because there was a large difference in performance between running Deus Ex: Mankind Divided with DX11 and DX12 with the Ryzen CPU, I included both the results in this graph. However, I only included DX12 results for Total War: Warhammer and not the DX11 results since the R5 1600 was much faster in both tests. I favored DX12 as it's the newer API.
If you care at all about value, the Ryzen 5 1600 is clearly the way to go. This is why we recently named it the best value performance desktop CPU. It was unlikely that the Core i7-7800X was going to change that, but we hoped the performance would at least be a compelling reason to buy Intel's new six-core processor.
Ryzen will hit 4GHz with the box cooler but it will be a more mild experience with a $20 aftermarket cooler like the Cooler Master 212, so keep that in mind. The 7800X on the other hand cannot be overclocked to 4.7GHz using a 240mm AIO closed loop solution. Instead, it required a $380 custom loop setup to achieve that result."
https://www.techspot.com/review/1450-core-i7-vs-ryzen-5-hexa-core/page9.html

None of this is surprising; we predicted most of it.

Look at power consumption: We've known Ryzen was build on a LPP node for a while now, so it's not surprising it consumes less power at stock when compared to Intel. But also notice how clocked at 4GHz how Ryzen puts up similar numbers to the 7700k clocked at 4.9GHz? As we also speculated for months, there's a point where power draw starts to increase rapidly, again due to the choice of node. Intel CPUs are going to scale much better power wise with increasing clocks, which also plays into it's (slight) performance edge over Ryzen in gaming.

Now for game performance: We've speculated months before Ryzen even launched that IPC would be around Haswell, which was pretty much spot on. Knowing that, and knowing that Haswell isn't significantly CPU bottlenecked, one would expect Ryzen ~ i7 level performance in gaming. Which is pretty much what we see. Again, nothing surprising.

Ryzen is pretty much what I'd expect honestly. I said it from the beginning: If AMD can get it's IPC around Haswell levels, then it will be competitive with Intel chips on performance.

 

[cdrkf]

AMD Ryzen 5 1600 vs Intel Core i7-7800X: 30 Game Battle!
Page 9 : Power Consumption & The Verdict
By Steven Walton on July 21, 2017
Power Consumption & The Verdict

"I have to admit, this just seems wrong. I still can't get used to seeing AMD CPUs consuming less power than their Intel counterpart. It feels unnatural. Those power-hungry FX chips sat atop our power charts for so long and I'm glad we can finally move on from them.
For what seemed like a similar level of performance (we'll get to that in a moment), the R5 1600 consumed quite a lot less power compared to the 7800X. With both chips overclocked, the Ryzen system consumed 15% less power and 18% less before we did any overclocking -- a definitive win for AMD. Who predicted this at the beginning of the year? I'll admit it certainly wasn't me.
We tried to get to the point quickly on this one but there's no easy way to do that when benchmarking 30 PC games. Overall, we're impressed with the performance displayed by the Ryzen 5 1600. Granted, it was a good chip from day one but it seems like the few months of optimizations have made it even better.
Before we get to our great graph of averages, it's worth noting that after spending an extra two days confirming these results (mostly re-testing and comparing the Core i7-7700K and Ryzen 5 1600), we can say that while these chips are strangely close in performance, our figures are accurate.
Out of the box the R5 1600 is 13% slower than the more expensive 7700K. Keep in mind, there were a few GPU-limited games and even worse, a few that were frame-capped, they weren't limited to 60fps. Many had 120fps+ limits, so the Ryzen CPU was still pushing the GTX 1080 Ti hard."

"Once overclocked, the $215 R5 1600 was just 9% slower than the 7700K, while it matched the $415 7800X. I'm sure you're itching to see what the overall picture looks like so we won't wait any longer.
Just looking at the 7800X and R5 1600, here are the overclocked results when comparing the minimum frame rates. As we just saw, both averaged a minimum of 103fps across the 30 games tested. We can clearly see where the R5 1600 enjoyed some wins and suffered through a few loses."

"Some readers will undoubtedly declare that I'm biased towards a certain company, but if that was the case, there's no way you would see Deus Ex: Mankind Divided and Gears of War 4 in this list. Meanwhile, if I were biased towards Intel, I'd drop Total War: Warhammer and Civilization IV.
It's worth noting that the number of titles won and lost are equal, and even if we were to remove some of the worst performing games for either platform, it would only skew the results slightly toward either camp, which is the beauty of testing with such a massive sample of games.
For example, if we remove AMD's worst performing title, Gears of War 4, the R5 1600 would become just 1% faster than the 7800X. Remove AMD's second worst title, Deus Ex Mankind Divided using DX12, and the R5 1600 is still just 1% faster on average.
Please note that because there was a large difference in performance between running Deus Ex: Mankind Divided with DX11 and DX12 with the Ryzen CPU, I included both the results in this graph. However, I only included DX12 results for Total War: Warhammer and not the DX11 results since the R5 1600 was much faster in both tests. I favored DX12 as it's the newer API.
If you care at all about value, the Ryzen 5 1600 is clearly the way to go. This is why we recently named it the best value performance desktop CPU. It was unlikely that the Core i7-7800X was going to change that, but we hoped the performance would at least be a compelling reason to buy Intel's new six-core processor.
Ryzen will hit 4GHz with the box cooler but it will be a more mild experience with a $20 aftermarket cooler like the Cooler Master 212, so keep that in mind. The 7800X on the other hand cannot be overclocked to 4.7GHz using a 240mm AIO closed loop solution. Instead, it required a $380 custom loop setup to achieve that result."
https://www.techspot.com/review/1450-core-i7-vs-ryzen-5-hexa-core/page9.html

None of this is surprising; we predicted most of it.

Look at power consumption: We've known Ryzen was build on a LPP node for a while now, so it's not surprising it consumes less power at stock when compared to Intel. But also notice how clocked at 4GHz how Ryzen puts up similar numbers to the 7700k clocked at 4.9GHz? As we also speculated for months, there's a point where power draw starts to increase rapidly, again due to the choice of node. Intel CPUs are going to scale much better power wise with increasing clocks, which also plays into it's (slight) performance edge over Ryzen in gaming.

Now for game performance: We've speculated months before Ryzen even launched that IPC would be around Haswell, which was pretty much spot on. Knowing that, and knowing that Haswell isn't significantly CPU bottlenecked, one would expect Ryzen ~ i7 level performance in gaming. Which is pretty much what we see. Again, nothing surprising.

Ryzen is pretty much what I'd expect honestly. I said it from the beginning: If AMD can get it's IPC around Haswell levels, then it will be competitive with Intel chips on performance.

All true (at least for those of us being realistic)- what I would say though, is Ryzen does present quite an interesting option now when looking at a gaming machine.

Currently no one would argue, if maximum performance in games is your goal, the 7700k is the fastest chip. What I wonder though is would a Ryzen 6 or 8 core chip be a better long term investment. I know the argument (games don't use that many threads), however I've been gaming long enough to remember 'you'll only ever need a single core' changing to 'you'll NEVER need more than a dual core for games' to 'Quad core is all that's needed'.... My guess would be in the next few years (especially in light on Intel's move to hex core on the mainstream parts for Coffee lake) that 6 core / 12 thread will become the optimum cpu that studios are designing to.

That means that- for the short term the 7700k won't be challenged, but if you're like me and you like to keep a system for 5+ years Ryzen might actually be the better choice. It's like how Phenom II X6 can still run modern games (ok it struggles with the latest titles but that is a *very* old CPU now). The same can be said of the old i7 2600k which still looks strong with a good overclock on it (whereas the 4 thread 2500k struggles). Interesting times!

 

[gamerk316]

Currently no one would argue, if maximum performance in games is your goal, the 7700k is the fastest chip. What I wonder though is would a Ryzen 6 or 8 core chip be a better long term investment. I know the argument (games don't use that many threads), however I've been gaming long enough to remember 'you'll only ever need a single core' changing to 'you'll NEVER need more than a dual core for games' to 'Quad core is all that's needed'.... My guess would be in the next few years (especially in light on Intel's move to hex core on the mainstream parts for Coffee lake) that 6 core / 12 thread will become the optimum cpu that studios are designing to.

That means that- for the short term the 7700k won't be challenged, but if you're like me and you like to keep a system for 5+ years Ryzen might actually be the better choice. It's like how Phenom II X6 can still run modern games (ok it struggles with the latest titles but that is a *very* old CPU now). The same can be said of the old i7 2600k which still looks strong with a good overclock on it (whereas the 4 thread 2500k struggles). Interesting times!

Let me address this specific point.

Going to dual core was always going to help performance, regardless of how a game is coded, for the simple reason you don't stall out your application whenever the OS decides to run one of its several hundred (if not thousand) threads for all the background operations a modern OS is expected to perform. That's one reason why even purely single threaded applications see at least some benefit from a dual core system; you're getting the OS out of the way.

Going to four doesn't help games as much, but you then have to consider the somewhat massive GPU render thread that existed prior to recent rendering APIs [DX12/Vulkan]. So you still have benefit: One core for the main game engine, one core for the rendering engine, and two cores to handle background/OS operations plus whatever low-workload threads the app in question has. So four cores becomes the sweet spot.

Now, the afore-mentioned GPU rendering thread can now be broken up into smaller chunks via new rendering APIs. While this doesn't by itself have any positive performance benefit, it does allow for lower-performance but many-core CPUs to perform the same as fewer-core but higher-performance CPUs. Essentially, while the total workload to perform is still the same, it can be handled in smaller chunks. In this type of case, a CPU architecture like Bulldozer would be less disadvantaged like it would have been in the past (much less chance of a CPU bottleneck on one core while other cores are idle), but this change in style by itself won't lead to any additional performance. Basically: If no CPU bottleneck exists, using more threads won't add any performance.

Which gets us to the crux of the issue here: Games are not significantly affected by CPU performance. Hell, half of the total CPU time for games nowadays is managing the GPU driver, and that's been broken up now. There literally isn't a lot games can do independent of the rendering engine that would allow them to eat that much CPU resources.

So what I'm expecting is game performance to basically track with single-core performance, which higher clocked chips and more optimized processors performing faster then previous ones, with core counts much beyond four continuing to have negligible performance impacts going forward.

This is a repeat of the same exact argument people made for the 8150/8350: Why purchase a 2600k/3770k when, some undefined time from now, the 8150/8350 could potentially perform better? And I'm railing against that reasoning now just like I did then, and guess what? Even with games more threadable now, and even with the GPU thread broken up, the 8350 still doesn't compare to even a stock 2600k at gaming. Why? Because when no individual CPU core is bottlenecked, relative single-core performance dominates performance. Adding more cores at that point just wastes resources.

 

[jdwii]

^^^ 2500K did age better but at that time bulldozer was MUCH slower on a core to core basis then Ryzen is to skylake. Also its important to note that the 4 core 4 threaded 2500K is starting to age quite a bit now where the 2600K is keeping up better. That is over the extra threads. 8 core piledriver on a per core basis is very slow even at 5ghz. Actually as a side project i was going to down clock my 8 core ryzen to try and match 8 core 5ghz PD.

Already know 3.2ghz is to high based on my benchmarks with Cinebench and Dolphin as well as a few others.

We are starting to see games use more then 4 cores today that is why a 6900K at stock is even with a 7700K at 5ghz in some games like watch dogs 2.

http://www.legitreviews.com/wp-content/uploads/2017/03/ryzen-dolphin-1800x-2.jpg

 

[daerohn]

Currently no one would argue, if maximum performance in games is your goal, the 7700k is the fastest chip. What I wonder though is would a Ryzen 6 or 8 core chip be a better long term investment. I know the argument (games don't use that many threads), however I've been gaming long enough to remember 'you'll only ever need a single core' changing to 'you'll NEVER need more than a dual core for games' to 'Quad core is all that's needed'.... My guess would be in the next few years (especially in light on Intel's move to hex core on the mainstream parts for Coffee lake) that 6 core / 12 thread will become the optimum cpu that studios are designing to.

That means that- for the short term the 7700k won't be challenged, but if you're like me and you like to keep a system for 5+ years Ryzen might actually be the better choice. It's like how Phenom II X6 can still run modern games (ok it struggles with the latest titles but that is a *very* old CPU now). The same can be said of the old i7 2600k which still looks strong with a good overclock on it (whereas the 4 thread 2500k struggles). Interesting times!

Let me address this specific point.

Going to dual core was always going to help performance, regardless of how a game is coded, for the simple reason you don't stall out your application whenever the OS decides to run one of its several hundred (if not thousand) threads for all the background operations a modern OS is expected to perform. That's one reason why even purely single threaded applications see at least some benefit from a dual core system; you're getting the OS out of the way.

Going to four doesn't help games as much, but you then have to consider the somewhat massive GPU render thread that existed prior to recent rendering APIs [DX12/Vulkan]. So you still have benefit: One core for the main game engine, one core for the rendering engine, and two cores to handle background/OS operations plus whatever low-workload threads the app in question has. So four cores becomes the sweet spot.

Now, the afore-mentioned GPU rendering thread can now be broken up into smaller chunks via new rendering APIs. While this doesn't by itself have any positive performance benefit, it does allow for lower-performance but many-core CPUs to perform the same as fewer-core but higher-performance CPUs. Essentially, while the total workload to perform is still the same, it can be handled in smaller chunks. In this type of case, a CPU architecture like Bulldozer would be less disadvantaged like it would have been in the past (much less chance of a CPU bottleneck on one core while other cores are idle), but this change in style by itself won't lead to any additional performance. Basically: If no CPU bottleneck exists, using more threads won't add any performance.

Which gets us to the crux of the issue here: Games are not significantly affected by CPU performance. Hell, half of the total CPU time for games nowadays is managing the GPU driver, and that's been broken up now. There literally isn't a lot games can do independent of the rendering engine that would allow them to eat that much CPU resources.

So what I'm expecting is game performance to basically track with single-core performance, which higher clocked chips and more optimized processors performing faster then previous ones, with core counts much beyond four continuing to have negligible performance impacts going forward.

This is a repeat of the same exact argument people made for the 8150/8350: Why purchase a 2600k/3770k when, some undefined time from now, the 8150/8350 could potentially perform better? And I'm railing against that reasoning now just like I did then, and guess what? Even with games more threadable now, and even with the GPU thread broken up, the 8350 still doesn't compare to even a stock 2600k at gaming. Why? Because when no individual CPU core is bottlenecked, relative single-core performance dominates performance. Adding more cores at that point just wastes resources.

Today issue is not the single core performace anymore. The real question you should be asking is "The more money I pay for Intel justifies the performance gain I will receive now and in the future" and if you are not using your computer solely for gaming you can ask another question "Would AMD have better performance per dollar for the other programs ı use". According to your replies you can choose whatever you want. However AMD is once again in "high performance chips" market.

 

[gamerk316]

Yeah, the point is that AMD is now a viable option; they weren't before the Ryzen launch.

As for the 2500/2600 debate, it's important to remember that those 10-15% architecture IPC gains add up after five or six refreshes. Currently, the 2500k has the problem of not being able to keep up with increased CPU workload in games. Even just a few years ago, the 2500k wasn't significantly bottlenecked; it could handle the workload assigned to each CPU core. But as games use more and more CPU power, the 2500k is starting to find it's in a situation where at least one core is being overworked, leading to a CPU bottleneck. The 2600k, by contrast, is more robust as some of that work can be assigned to a HTT core, preventing said bottleneck. Ironically, a move toward smaller threads would actually help the 2500k somewhat going forward. The lower the amount of work a thread does, the better it can be scheduled to avoid a particular core from being bottlenecked.

 

[cdrkf]

Yeah, the point is that AMD is now a viable option; they weren't before the Ryzen launch.

As for the 2500/2600 debate, it's important to remember that those 10-15% architecture IPC gains add up after five or six refreshes. Currently, the 2500k has the problem of not being able to keep up with increased CPU workload in games. Even just a few years ago, the 2500k wasn't significantly bottlenecked; it could handle the workload assigned to each CPU core. But as games use more and more CPU power, the 2500k is starting to find it's in a situation where at least one core is being overworked, leading to a CPU bottleneck. The 2600k, by contrast, is more robust as some of that work can be assigned to a HTT core, preventing said bottleneck. Ironically, a move toward smaller threads would actually help the 2500k somewhat going forward. The lower the amount of work a thread does, the better it can be scheduled to avoid a particular core from being bottlenecked.

I think you are underestimating how CPU dependent many modern titles are- and how many threads they need.

It's demonstrable that 8 threads are better than 4 on the same uarch in quite a few titles. This all stems from the fact the simulations are getting more complex. You don't necessarily need that many threads to feed the gpu- but when you start animating and controlling large numbers of AI, separate things like audio into independent threads and include complex physics simulation then you see that 8 threads > 4 threads > 2 hardware + 2 virtual threads. If you're argument of 'ipc + clock speed' is all you need held up, the i3 7350k would be the best gaming cpu. It's not- a quad core i5 is better, and a quad core 8 thread i7 is better still.

The whole argument over PD becoming better than Sandy is a moot point because the IPC deficit was so large. Still I maintain that whilst a 7700k is the fastest gaming cpu *now*, I wonder if 8 threads might start to hold things back *in some games* (note if when you say game you mean 'shooter' well probably not, those still run fine on most dual cores- when I talk games I'm thinking things like RTS titles, AOTS, PA and upcoming stuff like Star Citizen which is having a significant engine rewrite to use as many cores as you can throw at it). It's worth noting the difference now between a R7 1700 (with an overclock on it) to the 7700k (also overclocked) isn't huge, you're not giving up much. I honestly think the 1700 is the better buy long term, if you're likely to be playing these newer, more heavily threaded titles. If you're more into lighter workloads (e.g. most e-sports games) then sure, the IPC advantage on the 7700k may well be more important. Modern games though do *hammer* cpu's- this argument of 'it's all the gpu' just isn't true- at least for the stuff I play (PA for example can thrash an 8 thread cpu, use as much memory as you can give it but will run fine on a mid range gpu).

 

[MrN1ce9uy]

Ryzen 3 1300X vs Core i5-7500 - http://cpu.userbenchmark.com/Compare/Intel-Core-i5-7500-vs-AMD-Ryzen-3-1300X/3648vs3930

Is Ryzen 3 on par with Kaby Lake's IPC? I suppose it could be that (being 2 user benchmarks run w/ the 1300X) one was at stock and the other overclocked. That best/worst bench counter doesn't always work correctly.

 

[goldstone77]

For more benchmarks follow the link
AMD Ryzen 3 Review
Ryzen Now Starts at $110
By Steven Walton on July 27, 2017

https://www.techspot.com/review/1455-ryzen-3/

 

[-Fran-]

And some APU news! Finally!

http://www.tomshardware.com/news/amd-apu-bristol-ridge-a-series-athlon,35096.html

Cheers!

 

[rgd1101]

And some APU news! Finally!

http://www.tomshardware.com/news/amd-apu-bristol-ridge-a-series-athlon,35096.html

Cheers!

That the old gen not zen.

 

[-Fran-]

That the old gen not zen.

Ah, bummer. You're right.

I got happy for a second there 8(

Cheers!

 

[MrN1ce9uy]

Well my expectations for Ryzen 3 have been exceeded. Good for AMD. Now the price seems fitting.

 

[gamerk316]

Yeah, the point is that AMD is now a viable option; they weren't before the Ryzen launch.

As for the 2500/2600 debate, it's important to remember that those 10-15% architecture IPC gains add up after five or six refreshes. Currently, the 2500k has the problem of not being able to keep up with increased CPU workload in games. Even just a few years ago, the 2500k wasn't significantly bottlenecked; it could handle the workload assigned to each CPU core. But as games use more and more CPU power, the 2500k is starting to find it's in a situation where at least one core is being overworked, leading to a CPU bottleneck. The 2600k, by contrast, is more robust as some of that work can be assigned to a HTT core, preventing said bottleneck. Ironically, a move toward smaller threads would actually help the 2500k somewhat going forward. The lower the amount of work a thread does, the better it can be scheduled to avoid a particular core from being bottlenecked.

I think you are underestimating how CPU dependent many modern titles are- and how many threads they need.

It's demonstrable that 8 threads are better than 4 on the same uarch in quite a few titles. This all stems from the fact the simulations are getting more complex. You don't necessarily need that many threads to feed the gpu- but when you start animating and controlling large numbers of AI, separate things like audio into independent threads and include complex physics simulation then you see that 8 threads > 4 threads > 2 hardware + 2 virtual threads. If you're argument of 'ipc + clock speed' is all you need held up, the i3 7350k would be the best gaming cpu. It's not- a quad core i5 is better, and a quad core 8 thread i7 is better still.

The whole argument over PD becoming better than Sandy is a moot point because the IPC deficit was so large. Still I maintain that whilst a 7700k is the fastest gaming cpu *now*, I wonder if 8 threads might start to hold things back *in some games* (note if when you say game you mean 'shooter' well probably not, those still run fine on most dual cores- when I talk games I'm thinking things like RTS titles, AOTS, PA and upcoming stuff like Star Citizen which is having a significant engine rewrite to use as many cores as you can throw at it). It's worth noting the difference now between a R7 1700 (with an overclock on it) to the 7700k (also overclocked) isn't huge, you're not giving up much. I honestly think the 1700 is the better buy long term, if you're likely to be playing these newer, more heavily threaded titles. If you're more into lighter workloads (e.g. most e-sports games) then sure, the IPC advantage on the 7700k may well be more important. Modern games though do *hammer* cpu's- this argument of 'it's all the gpu' just isn't true- at least for the stuff I play (PA for example can thrash an 8 thread cpu, use as much memory as you can give it but will run fine on a mid range gpu).

Don't confuse the number of threads a program uses with how many cores get loaded by significant (measurable) work. You're mixing and matching the two. Ignoring scheduling overheads, two threads doing X work each will perform the same as one thread doing 2X work. Creating more threads isn't creating a higher workload, it's breaking it up into smaller chunks.

The advantage of threading is you more fine grain CPU load across CPU cores; there's a much lower chance of a single CPU core being overloaded as your threads do less work on average. The disadvantage is a slight overall performance loss.

As for game workload, with the notable exception of Strategy games [especially Real-Time Strategy games], the majority of the work being done is performed by the rendering engine, not the actual "game" logic. And yes, I've measured it. The rendering process is a very large chunk of CPU time, simply because of the GPU driver layer.

 

[naturesninja]

I don't believe either of those numbers on Userbenchmark are overclocked. I think those are just two samples. Very good scores, as Userbenchmark isn't the end all be all, it does pretty accurately place components where they should be.

One thing that I find interesting is the single/quad core performance of the 1300x exceeds that of the R5 1400 in both OC'd and non OC'd (1300X shows 3.5/3.6GHz, 1400 shows 4.0GHz). This leads me to believe the 1300x is possibly just using one CCX instead of 2 cores on each 2 CCXs. That, or the poster of said benches has access to some new mc/bios that isn't available yet to the public. Though I find it a little odd that a 1080ti was used in conjunction with the peak OC bench if AMD posted it (or maybe not...).