Discussion AMD Ryzen MegaThread! FAQ and Resources

[8350rocks]

What about the 250ms hang in frame times from the 7700K that his graph showed? The 5960X also experienced a hang that was 180ms.

Considering that Ryzen never broke 30ms gap, and at no point was considered stuttering, I would think the hard numbers from the frame time graph would corroborate the user experience. Especially true if we are considering it was duplicated by multiple individuals playing the same games across the same systems....

There is your definitive proof that Ryzen was smoother than Intel. Frame times do not lie, and a 180-250ms hang in frame times would be a horrendously bad stutter with a noticeable effect on performance.

I'm going to point out that normal tech review sites have found absolutely no evidence of those types of hangs, especially when you consider that a 250ms hang time represents 16 frames being consecutively dropped. Which would be quite noticeable, even without latency benchmarking.

So I'll come out and say it: the test is being rigged is some way to make Intel appear worse. Because if the 7700k (or other, slower, Intel CPUs such as my 2600k) had those types of issues in normal usage, someone would have noticed by now. Or are you going to accuse every single tech site in existence of being Intel shills for not reporting this issue?

Gamer, he has the fraps capture from the hang, and it was run during the video.

Seriously...the test is not being rigged. He even comments that he would have thought Nvidia's drivers should have just dropped the single frame and moved on, it makes the most sense from his perspective. The drivers, however, did not do that. He checks signal interrupts and all sorts of things. If you develop on windows, he literally had EVERYTHING perfmon checks enabled to see what was going on the whole time.

 

[8350rocks]

Zen also scales better with overclock because increasing the CPU clock increases cache performance as well, which reduces cache latency. Not to mention IF scaling with RAM speed also increases performance with RAM OC or AMP/XMP profiles.

but the problem is memory here, overclocking the cpu will make memory problem more obvious and give minimum benifit in games or tasks that depend on memory more. And l1 cache operate at cpu speed, l2 cache also operate at cpu speed and close to cpu speed in some, so overclocking both intel and amd will overclock those by same amount thus maintaining same ratio of those component's clock.
i would believe it when i see it ( preferably in gta5 ) , keep ram at 2133mhz and then overclock cpu only to see how much more fps we get, in my 1090t i get 30% performance increase from 1800mhz cl9 ram 2925mhz imc vs 1200mhz cl9 ram 2025mhz imc, on the other hand a cpu overclock from 3.6 to 4.05ghz gives only ~3% more fps ( at 1800mhz cl9 ram 2925mhz imc ) ( default setting 720p on 7950@1.3ghz )

Why would you buy a 7700K for ~$350, when you can buy a 12 thread R5-1600X that performs just as well in nearly everything?

if the want a gaming rig then its 7700k if they want more of transcoding/redering rig then it is 1600x, in games all zen performs more close to 2600k than to 7700k and 2600k overclock much higher % than ryzen and support great memory overclock too thus more to squeze from it.

What about the 250ms hang in frame times from the 7700K that his graph showed? The 5960X also experienced a hang that was 180ms.

pure bs, to get more click/views, post something normal then nobody cares so post something "omg you won't believe this" then then get more views ( just like a cancer to review media ) and fanboys will spread the link thus millions of views for bs.

Since skylake, Intel has divorced cache speed from CPU frequency.

Also, he never used a clickbait title, and even emphasized there were 3 specific instances in the benchmark where a single frame hung for an extended period.

He notes that in the 0.1% lows, he left them out as an outlier, because it would have tanked the minimum FPS for the 7700K, so he basically gives them a pass on it.

If anything, he is nearly pandering to intel on this one. Watch the video.

 

[goldstone77]

Full disclosure all 3 of my systems at home have Intel inside. That being said I'm excited for AMD this year for bringing competition back to the computer industry! Infinity Fabric uses RAM frequency as it's reference clock. Benchmarks not using high frequency low latency RAM are not going to take full advantage of the architecture. Also, there is some value that can be found with low latency RAM. 3200MHz CAS 14 Single Rank RAM being most compatible with Ryzen atm."Overall, going from CL16 at 2133Mhz to C14 at 3200 MHz increased my fps in Gears 4 by 25%"
https://community.amd.com/message/2790538#comment-2790538
https://community.amd.com/servlet/JiveServlet/showImage/2-2790538-119043/pastedImage_0.png
Sure the majority of people game at 1080p, but they also game at 60HZ! Most of difference between Ryzen and Intel gaming at this level is purely academic for the vast majority of users! While having the extra mult-threading and computing power will be sought after by many. That being said I'm gaming with a 2500K@4.5GHz, Samsung 850 EVO 500GB, 16GB 2400 CAS 10 RAM, and a RX 480 4GB, and don't have a problem playing any games at 1080p. And my processor is going on 7 years old now!
People are not making apple to apple comparisons otherwise you would see in games optimized for Ryzen vs. Intel show very similar results when paired with high frequency low latency RAM. Take a good look at FPS in games also optimized for Ryzen. IPC matters very little vs. games being optimized for Ryzen! Also, look at the process node differences. Note the advertised 14nm actually measures 17nm for AMD and 13nm for Intel. Ryzen actually has really good architecture considering the much smaller process node, and lack of being able to overclocking when compared to Intel. Intel's CEO said that they are having difficulties reducing the size of the process node, and if you look at the Node Process flow charts Intel is now pulling up the rear of the pack out of the 4 foundries. When Intel brings it's 10nm(9.5nm) to market mid 2018 AMD will drop it's 7nm(9.2nm), which will almost cut Ryzens current process node in half at 7.8nm vs. Intel's 3.5nm. This will be the first time AMD has had a smaller process node than Intel. Consider the great amount of performance gained and reduced energy cost from shrinking the process node! This will be HUGE for Ryzen. Ryzen is essentially the same chip 1800x, 1700x, 1700, 1600x,1600, 1500x, and 1500. All capable of a overclocking. 1700 for $319 at 3.9GHz is basically at 1800x. And the 1600 for $219 overclocked will offer really good value to performance! World wide not everyone can afford high dollar Intel chips and love the performance and price that Ryzen can pass on to it's customers by making 1 design fits all concept! This really is an exciting time for computing!

 

[randomizer]

It's amusing that two people perceive completely opposing deliberate bias in the same video. I guess it must be well balanced.

 

[jdwii]

Spoiler

XFR is 200MHz for the 1500X and that is for standard cooling. It could get higher clocks on an open case with low tempts, because the XFR turbo depends on temperature.

RyZen is not closing the gap significantly. Here you have the impact of the new AGESA update that addressed the memory latency issue, how you can check yourself the overall performance has not varied a bit

Here you have a recent graph with last data for RyZen including the new models. First games and below compute

If you compare the R5 1400 with the i7 2600k. You can easily deduce

RyZen IPC+SMT ~ Haswell
RyZen IPC ~ Sandy

which is the same conclusion obtained at launch day with the reviews of the 1800X.

Yeah... Well, No.

A single data point doesn't help compare "day1" to "current day". I still see in today's data that Ryzen is within 7% of Kaby Lake clock per clock when day1 was 12%-ish. Whatever the reasons, I don't really care, but like you say "the data is there". The most beaten to death horse, CB, seems to agree with me there.

Which, at the same time, makes me even more curious to explain *why* we see bigger differences elsewhere. Your default answer is "because latency". Well, again, no. That is not a convincing answer. Have more data that could indicate why? Something that you haven't thought of yet?

I have read that different RAM and cooling should be the biggest differentiation variables, but outside of Toms and Anands, I haven't read anything else around those topics.

Cheers!

Cinebench seems to really like Ryzen i can't figure out how it actually gets those scores as in almost every other benchmark besides CPU-Z Ryzen single core is between sandy-haswell.

 

[Nope 1151]

Spoiler

XFR is 200MHz for the 1500X and that is for standard cooling. It could get higher clocks on an open case with low tempts, because the XFR turbo depends on temperature.

RyZen is not closing the gap significantly. Here you have the impact of the new AGESA update that addressed the memory latency issue, how you can check yourself the overall performance has not varied a bit

Here you have a recent graph with last data for RyZen including the new models. First games and below compute

If you compare the R5 1400 with the i7 2600k. You can easily deduce

RyZen IPC+SMT ~ Haswell
RyZen IPC ~ Sandy

which is the same conclusion obtained at launch day with the reviews of the 1800X.

Yeah... Well, No.

A single data point doesn't help compare "day1" to "current day". I still see in today's data that Ryzen is within 7% of Kaby Lake clock per clock when day1 was 12%-ish. Whatever the reasons, I don't really care, but like you say "the data is there". The most beaten to death horse, CB, seems to agree with me there.

Which, at the same time, makes me even more curious to explain *why* we see bigger differences elsewhere. Your default answer is "because latency". Well, again, no. That is not a convincing answer. Have more data that could indicate why? Something that you haven't thought of yet?

I have read that different RAM and cooling should be the biggest differentiation variables, but outside of Toms and Anands, I haven't read anything else around those topics.

Cheers!

Cinebench seems to really like Ryzen i can't figure out how it actually gets those scores as in almost every other benchmark besides CPU-Z Ryzen single core is between sandy-haswell.

Could it be that Cbench is not as latency sensitive? Let's not start theories about "AMD BIAS" on compilers.

Edit: Is anyone seeing some price differences with the 1600? It is 290 USD here in China.

 

[8350rocks]

It's amusing that two people perceive completely opposing deliberate bias in the same video. I guess it must be well balanced.

Actually, what is amusing is the number of people perceiving bias towards intel from the video got so large a big reddit thread started to runaway. Wendell hopped in and clarified.

https://www.reddit.com/r/Amd/comments/659fna/double_standards_in_tech_journalism_can_you/dg8sabk/

I understand he took the diplomatic approach, and did not want to point the finger at anything specific because he was not sure. I can respect the fact that he stands by that, and simply provides the data they captured as to allow people to draw their own conclusions from the raw data.

He says that no one can tell the difference between them for 98% of game play...I think people should consider that a victory as it stands. The fact that there is reproducible game play effecting hangs from Intel CPUs using Nvidia drivers is just a foil to all the "My intel CPU never stutters, ever..." crowd.

 

[juanrga]

XFR is 200MHz for the 1500X and that is for standard cooling. It could get higher clocks on an open case with low tempts, because the XFR turbo depends on temperature.

RyZen is not closing the gap significantly. Here you have the impact of the new AGESA update that addressed the memory latency issue, how you can check yourself the overall performance has not varied a bit

Here you have a recent graph with last data for RyZen including the new models. First games and below compute

If you compare the R5 1400 with the i7 2600k. You can easily deduce

RyZen IPC+SMT ~ Haswell
RyZen IPC ~ Sandy

which is the same conclusion obtained at launch day with the reviews of the 1800X.

Yeah... Well, No.

A single data point doesn't help compare "day1" to "current day". I still see in today's data that Ryzen is within 7% of Kaby Lake clock per clock when day1 was 12%-ish. Whatever the reasons, I don't really care, but like you say "the data is there". The most beaten to death horse, CB, seems to agree with me there.

Which, at the same time, makes me even more curious to explain *why* we see bigger differences elsewhere. Your default answer is "because latency". Well, again, no. That is not a convincing answer. Have more data that could indicate why? Something that you haven't thought of yet?

I have read that different RAM and cooling should be the biggest differentiation variables, but outside of Toms and Anands, I haven't read anything else around those topics.

At launch RyZen was 10--20% behind Broadwell clock-for-clock. Improved BIOS, Windows updates, and the extra cache in R5 models can have closed the gap by about 5%. But if you check the above recent data 4C/8T RyZen performs more like a 4C/8T Sandy Bridge than like a 4C/8T Kabylake.

 

[juanrga]

Cinebench seems to really like Ryzen i can't figure out how it actually gets those scores as in almost every other benchmark besides CPU-Z Ryzen single core is between sandy-haswell.

Could it be that Cbench is not as latency sensitive? Let's not start theories about "AMD BIAS" on compilers.

L2 cache size. Both CineBench and CPU-Z are very sensitive to L2 cache. RyZen shines on both benches thanks to 512KB of L2. Intel chips have 256KB

Skylake-X updates L2 cache to 1MB per core. You will see how well Skylake-X performs on CineBench and CPU-Z. We could even see 15% IPC gains here compared to Broadwell. Early rumors mention that 8-core Skylake-X beats 10-core Broadwell on CineBench.

 

[randomizer]

Actually, what is amusing is the number of people perceiving bias towards intel from the video got so large a big reddit thread started to runaway.

People's perception of reality is at least as far from reality as reality TV is.

 

[Nope 1151]

Cinebench seems to really like Ryzen i can't figure out how it actually gets those scores as in almost every other benchmark besides CPU-Z Ryzen single core is between sandy-haswell.

Could it be that Cbench is not as latency sensitive? Let's not start theories about "AMD BIAS" on compilers.

L2 cache size. Both CineBench and CPU-Z are very sensitive to L2 cache. RyZen shines on both benches thanks to 512KB of L2. Intel chips have 256KB

Skylake-X updates L2 cache to 1MB per core. You will see how well Skylake-X performs on CineBench and CPU-Z. We could even see 15% IPC gains here compared to Broadwell. Early rumors mention that 8-core Skylake-X beats 10-core Broadwell on CineBench.

Hmm... Do you think it is possible to use HBM on L2, L1 dies? If so, are there any benefits?

 

[TMTOWTSAC]

Cinebench seems to really like Ryzen i can't figure out how it actually gets those scores as in almost every other benchmark besides CPU-Z Ryzen single core is between sandy-haswell.

Could it be that Cbench is not as latency sensitive? Let's not start theories about "AMD BIAS" on compilers.

L2 cache size. Both CineBench and CPU-Z are very sensitive to L2 cache. RyZen shines on both benches thanks to 512KB of L2. Intel chips have 256KB

Skylake-X updates L2 cache to 1MB per core. You will see how well Skylake-X performs on CineBench and CPU-Z. We could even see 15% IPC gains here compared to Broadwell. Early rumors mention that 8-core Skylake-X beats 10-core Broadwell on CineBench.

Hmm... Do you think it is possible to use HBM on L2, L1 dies? If so, are there any benefits?

As I understand it, HBM is primarily about bandwidth, ie how much can be transferred, while caches are primarily about speed, ie how quickly can something be transferred.

 

[8350rocks]

Actually, what is amusing is the number of people perceiving bias towards intel from the video got so large a big reddit thread started to runaway.

People's perception of reality is at least as far from reality as reality TV is.

While true, you do realize that works both ways...correct?

 

[juanrga]

Cinebench seems to really like Ryzen i can't figure out how it actually gets those scores as in almost every other benchmark besides CPU-Z Ryzen single core is between sandy-haswell.

Could it be that Cbench is not as latency sensitive? Let's not start theories about "AMD BIAS" on compilers.

L2 cache size. Both CineBench and CPU-Z are very sensitive to L2 cache. RyZen shines on both benches thanks to 512KB of L2. Intel chips have 256KB

Skylake-X updates L2 cache to 1MB per core. You will see how well Skylake-X performs on CineBench and CPU-Z. We could even see 15% IPC gains here compared to Broadwell. Early rumors mention that 8-core Skylake-X beats 10-core Broadwell on CineBench.

Hmm... Do you think it is possible to use HBM on L2, L1 dies? If so, are there any benefits?

Zero benefits. HBM uses DRAM which optimizes density over speed. L1/L2 caches use SRAM which optimizes speed over density. Moreover,

 

[juanrga]

At launch RyZen was 10--20% behind Broadwell clock-for-clock. Improved BIOS, Windows updates, and the extra cache in R5 models can have closed the gap by about 5%. But if you check the above recent data 4C/8T RyZen performs more like a 4C/8T Sandy Bridge than like a 4C/8T Kabylake.

There were no benchmarks showing any such deficit against Broadwell.

I am going to ask you to provide proof of your claims from a reputable source.

At contrary, there are several dozens of benches showing that, and all them from mainstream top sources. The benches have been provided here and you even commented about them then. The mentioned 10--20% gap are averages from twenty compute and game measurements. Since they are averages, the gap can be much higher on some special case. I am not going to review all the benches to check which is the highest gap between Zen and Broadwell, but I remember perfectly the next compute benches, where Zen is from 26% to 31% slower than Broadwell clock-for-clock

And I remember this game where RyZen was 62% slower than Broadwell clock for clock.

Although to be fair that was before the game was patched.

 

[goldstone77]

At launch RyZen was 10--20% behind Broadwell clock-for-clock. Improved BIOS, Windows updates, and the extra cache in R5 models can have closed the gap by about 5%. But if you check the above recent data 4C/8T RyZen performs more like a 4C/8T Sandy Bridge than like a 4C/8T Kabylake.

There were no benchmarks showing any such deficit against Broadwell.

I am going to ask you to provide proof of your claims from a reputable source.

At contrary, there are several dozens of benches showing that, and all them from mainstream top sources. The benches have been provided here and you even commented about them then. The mentioned 10--20% gap are averages from twenty compute and game measurements. Since they are averages, the gap can be much higher on some special case. I am not going to review all the benches to check which is the highest gap between Zen and Broadwell, but I remember perfectly the next compute benches, where Zen is from 26% to 31% slower than Broadwell clock-for-clock

And I remember this game where RyZen was 62% slower than Broadwell clock for clock.

Although to be fair that was before the game was patched.

Benchmarks not using high frequency low latency RAM are not going to take full advantage of the architecture. Also, there is some value that can be found with low latency RAM. 3200MHz CAS 14 Single Rank RAM being most compatible with Ryzen atm."Overall, going from CL16 at 2133Mhz to C14 at 3200 MHz increased my fps in Gears 4 by 25%"
https://community.amd.com/message/2790538#comment-27905...
https://community.amd.com/servlet/JiveServlet/showImage...

 

[juanrga]

Benches used stock settings for memory. Overclocking RAM favors both Intel and AMD, as has been demonstrated again and again in this thread. OC RAM is not a magic bullet that will reduce a ~15% microarchitectural gap to 0%. I am not going to copy and paste all the graphs again. I will just re-add two of the more favorable cases: one game and one compute

Also I got info from a RyZen user that just installed the new AGESA that supposedly fixes RAM latency on RyZen. Yes, the new AGESA reduces latency by about 6 nanoseconds, but he lost OC capacity and is now running back RAM at 2400MHz. Any minor improvement on same clocks latency has been lost due to reduction in frequency. In his own words:

If anything, the Agesa 1004a update caused me problems. It killed my RAM overclock. Have to run at 2400 now. So AMD is moving in the wrong direction with optimizations and fixes.

 

[randomizer]

While true, you do realize that works both ways...correct?

I'm not sure what "ways" you're referring to.

 

[goldstone77]

Benches used stock settings for memory. Overclocking RAM favors both Intel and AMD, as has been demonstrated again and again in this thread. OC RAM is not a magic bullet that will reduce a ~15% microarchitectural gap to 0%. I am not going to copy and paste all the graphs again. I will just re-add two of the more favorable cases: one game and one compute

Also I got info from a RyZen user that just installed the new AGESA that supposedly fixes RAM latency on RyZen. Yes, the new AGESA reduces latency by about 6 nanoseconds, but he lost OC capacity and is now running back RAM at 2400MHz. Any minor improvement on same clocks latency has been lost due to reduction in frequency. In his own words:

If anything, the Agesa 1004a update caused me problems. It killed my RAM overclock. Have to run at 2400 now. So AMD is moving in the wrong direction with optimizations and fixes.

Can you use valid sources? I'm not sure where those foreign benchmarks are coming from. And no idea what their test setups were. RAM directly affects Infinity Fabric, and has a much greater affect on system performance than RAM does on Intel! And each users experience with Ryzen will vary with each system motherboard, updates, etc!
"Interestingly, the gains from 2667MHz to 3200MHz were approximately double those from 2133MHz to 2667MHz, despite an identical 533MHz increase. Does this mean increasing returns as the frequency gets higher?" Intel doesn't get a 10% performance increase going from 2667MHz to 3200MHz! https://community.amd.com/servlet/JiveServlet/showImage/2-2790040-118998/Bench.PNG

 

[juanrga]

Benches used stock settings for memory. Overclocking RAM favors both Intel and AMD, as has been demonstrated again and again in this thread. OC RAM is not a magic bullet that will reduce a ~15% microarchitectural gap to 0%. I am not going to copy and paste all the graphs again. I will just re-add two of the more favorable cases: one game and one compute

Also I got info from a RyZen user that just installed the new AGESA that supposedly fixes RAM latency on RyZen. Yes, the new AGESA reduces latency by about 6 nanoseconds, but he lost OC capacity and is now running back RAM at 2400MHz. Any minor improvement on same clocks latency has been lost due to reduction in frequency. In his own words:

If anything, the Agesa 1004a update caused me problems. It killed my RAM overclock. Have to run at 2400 now. So AMD is moving in the wrong direction with optimizations and fixes.

Can you use valid sources? I'm not sure where those foreign benchmarks are coming from. And no idea what their test setups were. RAM directly affects Infinity Fabric, and has a much greater affect on system performance than RAM does on Intel! And each users experience with Ryzen will vary with each system motherboard, updates, etc!

Uh? I am using Hardware.fr because they have provided the most extensive review of Zen. They are a well-known top site and they have tested virtually anything about RyZen: IPC, latency, throughput, BIOS updates, CCX-CCX interconnect, SMT, core parking, scheduler, power management policies,... and it is simpler to me to go their site and take anything that I need, instead goggling a dozen of reviews to check what site has the info I need, like this bench showing how Skylake improves performance on ARMA3 with faster RAM

RAM affect directly Infinity Fabric and this is only relevant if you are using a memory-bound workload. And memory bound workloads also run faster on Intel chips when the RAM is overclocked.

 

[goldstone77]

Benches used stock settings for memory. Overclocking RAM favors both Intel and AMD, as has been demonstrated again and again in this thread. OC RAM is not a magic bullet that will reduce a ~15% microarchitectural gap to 0%. I am not going to copy and paste all the graphs again. I will just re-add two of the more favorable cases: one game and one compute

Also I got info from a RyZen user that just installed the new AGESA that supposedly fixes RAM latency on RyZen. Yes, the new AGESA reduces latency by about 6 nanoseconds, but he lost OC capacity and is now running back RAM at 2400MHz. Any minor improvement on same clocks latency has been lost due to reduction in frequency. In his own words:

If anything, the Agesa 1004a update caused me problems. It killed my RAM overclock. Have to run at 2400 now. So AMD is moving in the wrong direction with optimizations and fixes.

Can you use valid sources? I'm not sure where those foreign benchmarks are coming from. And no idea what their test setups were. RAM directly affects Infinity Fabric, and has a much greater affect on system performance than RAM does on Intel! And each users experience with Ryzen will vary with each system motherboard, updates, etc!

Uh? I am using Hardware.fr because they have provided the most extensive review of Zen. They are a well-known top site and they have tested virtually anything about RyZen: IPC, latency, throughput, BIOS updates, CCX-CCX interconnect, SMT, core parking, scheduler, power management policies,... and it is simpler to me to go their site and take anything that I need, instead goggling a dozen of reviews to check what site has the info I need, like this bench showing how Skylake improves performance on ARMA3 with faster RAM

[/img]

RAM affect directly Infinity Fabric and this is only relevant if you are using a memory-bound workload. And memory bound workloads also run faster on Intel chips when the RAM is overclocked.

"Interestingly, the gains from 2667MHz to 3200MHz were approximately double those from 2133MHz to 2667MHz, despite an identical 533MHz increase. Does this mean increasing returns as the frequency gets higher?" Intel doesn't get a 10% performance increase going from 2667MHz to 3200MHz! https://community.amd.com/servlet/JiveServlet/showImage...

I can't read french, so there is no way I can validate any of the information you link from there.

 

[jdwii]

Spoiler

Benches used stock settings for memory. Overclocking RAM favors both Intel and AMD, as has been demonstrated again and again in this thread. OC RAM is not a magic bullet that will reduce a ~15% microarchitectural gap to 0%. I am not going to copy and paste all the graphs again. I will just re-add two of the more favorable cases: one game and one compute

Also I got info from a RyZen user that just installed the new AGESA that supposedly fixes RAM latency on RyZen. Yes, the new AGESA reduces latency by about 6 nanoseconds, but he lost OC capacity and is now running back RAM at 2400MHz. Any minor improvement on same clocks latency has been lost due to reduction in frequency. In his own words:

If anything, the Agesa 1004a update caused me problems. It killed my RAM overclock. Have to run at 2400 now. So AMD is moving in the wrong direction with optimizations and fixes.

Can you use valid sources? I'm not sure where those foreign benchmarks are coming from. And no idea what their test setups were. RAM directly affects Infinity Fabric, and has a much greater affect on system performance than RAM does on Intel! And each users experience with Ryzen will vary with each system motherboard, updates, etc!
"Interestingly, the gains from 2667MHz to 3200MHz were approximately double those from 2133MHz to 2667MHz, despite an identical 533MHz increase. Does this mean increasing returns as the frequency gets higher?" Intel doesn't get a 10% performance increase going from 2667MHz to 3200MHz! https://community.amd.com/servlet/JiveServlet/showImage/2-2790040-118998/Bench.PNG

Just so you know a poor memory controller and cheap manufacturing which is why the "Infinity Fabric" needs high frequency memory to compete more in line with Intel setups with slower speeds is NOT a good thing and its not a excuse, and its even worse that its hard to get Ryzen to even perform at those speeds as of today.

Intel processors also improve with higher frequency memory by a decent margin so any true not bias reviewer should test both processors at the same ram frequency.

Edit by cheap i mean smart but with a compromise

 

[goldstone77]

Spoiler

Benches used stock settings for memory. Overclocking RAM favors both Intel and AMD, as has been demonstrated again and again in this thread. OC RAM is not a magic bullet that will reduce a ~15% microarchitectural gap to 0%. I am not going to copy and paste all the graphs again. I will just re-add two of the more favorable cases: one game and one compute

Also I got info from a RyZen user that just installed the new AGESA that supposedly fixes RAM latency on RyZen. Yes, the new AGESA reduces latency by about 6 nanoseconds, but he lost OC capacity and is now running back RAM at 2400MHz. Any minor improvement on same clocks latency has been lost due to reduction in frequency. In his own words:

If anything, the Agesa 1004a update caused me problems. It killed my RAM overclock. Have to run at 2400 now. So AMD is moving in the wrong direction with optimizations and fixes.

Can you use valid sources? I'm not sure where those foreign benchmarks are coming from. And no idea what their test setups were. RAM directly affects Infinity Fabric, and has a much greater affect on system performance than RAM does on Intel! And each users experience with Ryzen will vary with each system motherboard, updates, etc!
"Interestingly, the gains from 2667MHz to 3200MHz were approximately double those from 2133MHz to 2667MHz, despite an identical 533MHz increase. Does this mean increasing returns as the frequency gets higher?" Intel doesn't get a 10% performance increase going from 2667MHz to 3200MHz! https://community.amd.com/servlet/JiveServlet/showImage/2-2790040-118998/Bench.PNG

Just so you know a poor memory controller and cheap manufacturing which is why the "Infinity Fabric" needs high frequency memory to compete more in line with Intel setups with slower speeds is NOT a good thing and its not a excuse, and its even worse that its hard to get Ryzen to even perform at those speeds as of today.

Intel processors also improve with higher frequency memory by a decent margin so any true not bias reviewer should test both processors at the same ram frequency.

Edit by cheap i mean smart but with a compromise

Infinity Fabric operates much like the FSB did, and that is why RAM affect performance the way it does.

 

[8350rocks]

While true, you do realize that works both ways...correct?

I'm not sure what "ways" you're referring to.

It works for everyone...not just a specific subset of people. So, in a manner, I suppose I am referring to over 7 billion ways. Myself, yourself, and this entire forum included.

 

[jdwii]

Spoiler

Benches used stock settings for memory. Overclocking RAM favors both Intel and AMD, as has been demonstrated again and again in this thread. OC RAM is not a magic bullet that will reduce a ~15% microarchitectural gap to 0%. I am not going to copy and paste all the graphs again. I will just re-add two of the more favorable cases: one game and one compute

Also I got info from a RyZen user that just installed the new AGESA that supposedly fixes RAM latency on RyZen. Yes, the new AGESA reduces latency by about 6 nanoseconds, but he lost OC capacity and is now running back RAM at 2400MHz. Any minor improvement on same clocks latency has been lost due to reduction in frequency. In his own words:

If anything, the Agesa 1004a update caused me problems. It killed my RAM overclock. Have to run at 2400 now. So AMD is moving in the wrong direction with optimizations and fixes.

Can you use valid sources? I'm not sure where those foreign benchmarks are coming from. And no idea what their test setups were. RAM directly affects Infinity Fabric, and has a much greater affect on system performance than RAM does on Intel! And each users experience with Ryzen will vary with each system motherboard, updates, etc!
"Interestingly, the gains from 2667MHz to 3200MHz were approximately double those from 2133MHz to 2667MHz, despite an identical 533MHz increase. Does this mean increasing returns as the frequency gets higher?" Intel doesn't get a 10% performance increase going from 2667MHz to 3200MHz! https://community.amd.com/servlet/JiveServlet/showImage/2-2790040-118998/Bench.PNG

Just so you know a poor memory controller and cheap manufacturing which is why the "Infinity Fabric" needs high frequency memory to compete more in line with Intel setups with slower speeds is NOT a good thing and its not a excuse, and its even worse that its hard to get Ryzen to even perform at those speeds as of today.

Intel processors also improve with higher frequency memory by a decent margin so any true not bias reviewer should test both processors at the same ram frequency.

Edit by cheap i mean smart but with a compromise

Infinity Fabric operates much like the FSB did, and that is why RAM affect performance the way it does.

Yes and that was all in well Amd's decision on how they made the architecture, not a good excuse.