The Definitive AMD Ryzen 7 Real-World Gaming Guide @ [H]

And AMD should be not? Catching up on Sandy in terms of gaming after all these years is not an achievement either.

Actually it is. Because Kaby isn't that far from Sandy either. And Ryzen is doing well, staying ahead of Sandy, despite a 500 MHz clock speed disadvantage compared to it. So even in gaming, the IPC of Ryzen is impressive for what it is. As the article tells us, if AMD can get the clockspeed up on Ryzen, it'll hang even with Kaby. New architecture and a new process? AMD did damned well, all things considered.

That being said, the new Intel HEDT offerings (at MUCH lower prices!) appear to be a salvo from Team Blue at Ryzen's status as the mixed-use CPU of choice. So AMD has their work cut out for them. They'll need some extra clockspeed from Zen in the near future, and some aggressive pricing from the new Threadripper line.

All to the good, though. So much more fun when I see Intel and AMD trading blows on this stuff. I missed the old days.
 
Did anyone bring up that ROTR got a DX12 patch specifically for Ryzen today? Curious how that plays out

Love these articles, can't wait for when they get memory to support to 3600, and they tackle this, there's quite a bit of potential upside to Ryzen with the game patches slowly trickling out.
 
I am not impressed with AMD, but I am also not impressed with Intel. I really wanted AMD to bring "it" to beat Intel over the head with some heavy handed competition. I didn't expect that to happen, but I was hopeful.

The gains over the last 6 years feel minuscule in comparison to mobile processor development. That seems to be where the real action is these days.

Thank you to the HardOCP team for the massive amount of work that went into this.
 
I am not impressed with AMD, but I am also not impressed with Intel. I really wanted AMD to bring "it" to beat Intel over the head with some heavy handed competition. I didn't expect that to happen, but I was hopeful.

The gains over the last 6 years feel minuscule in comparison to mobile processor development. That seems to be where the real action is these days.

Thank you to the HardOCP team for the massive amount of work that went into this.

I don't understand that viewpoint at all....in what other established sector do you see companies 1/10 the size even remotely able to compete on the same level, let alone be better in some segments?

The enthusiast gamer market is not by itself large enough to sell enough chips to make a multi-billion dollar investment prudent for a company AMDs size. Ryzen is a VERY smart play if you're a company that needs to throw all their weight behind one product design. The chip is price competitive across 95% of the product range and ideally suited to making price competitive move on the workstation front. Literally the only thing it really lags on is top end gaming performance. And how did it NOT give Intel some heavy-handed competition? Look at their response! Do you think Intel would be releasing the new HEDT chips RIGHT NOW if they weren't being pressured?
 
I'd really like to see a test where all cpus are clocked at 4ghz. It just feels weird to see a benchmark test where 1 cpu is clocked 1ghz higher and then claim it's faster in benchmarks than the one clocked at 4ghz. Second, I've shopped around for a new PC build and with the price difference between and AMD build and an Intel build, I can get a gtx 1060 with an Intel build but buy a GTX 1070 with my AMD build for just a few bucks more, depending on the card vendor.
 
I don't understand that viewpoint at all....in what other established sector do you see companies 1/10 the size even remotely able to compete on the same level, let alone be better in some segments?

The enthusiast gamer market is not by itself large enough to sell enough chips to make a multi-billion dollar investment prudent for a company AMDs size. Ryzen is a VERY smart play if you're a company that needs to throw all their weight behind one product design. The chip is price competitive across 95% of the product range and ideally suited to making price competitive move on the workstation front. Literally the only thing it really lags on is top end gaming performance. And how did it NOT give Intel some heavy-handed competition? Look at their response! Do you think Intel would be releasing the new HEDT chips RIGHT NOW if they weren't being pressured?

I never said anything about the size of the companies nor do I care frankly. This is the land of the giants where setting up the fabs and R&D runs you well into the billions with each iteration.

I agree with you to a certain extent, but AMD has yet to deliver something that provides enough of an incentive to switch from Intel in the server market, nor the gaming market. My reasoning lies mostly within the boundaries of cross cutting concerns of X86 overall as continual development like this impacts all areas of X86. Progress has slowed down due to technical considerations which is understandable, but I was hopeful for more progress in general from AMD to light a fire under Intel's ass.

I never said that I felt like AMD wasn't competitive with the recent processors. As someone who owns a 16c/32t V3 Xeon processor (and have for a while), I am not shocked to see that Intel had some HEDT chips lined up and ready to go once Ryzen dropped.
 
I'd really like to see a test where all cpus are clocked at 4ghz. It just feels weird to see a benchmark test where 1 cpu is clocked 1ghz higher and then claim it's faster in benchmarks than the one clocked at 4ghz. Second, I've shopped around for a new PC build and with the price difference between and AMD build and an Intel build, I can get a gtx 1060 with an Intel build but buy a GTX 1070 with my AMD build for just a few bucks more, depending on the card vendor.

That's like asking for a car with a gas engine and a diesel engine to be compared at the same RPM (good luck getting your diesel engine to 9000 RPM).
 
I'd really like to see a test where all cpus are clocked at 4ghz. It just feels weird to see a benchmark test where 1 cpu is clocked 1ghz higher and then claim it's faster in benchmarks than the one clocked at 4ghz. Second, I've shopped around for a new PC build and with the price difference between and AMD build and an Intel build, I can get a gtx 1060 with an Intel build but buy a GTX 1070 with my AMD build for just a few bucks more, depending on the card vendor.

I didn't really vet these sources out, but these seem to do the comparison you were interested in.


Basic conclusion - Ryzen is around 2% slower than kaby lake IPC wise and around 5% faster IPC than Broadwell-e

https://arstechnica.com/gadgets/2017/03/amd-ryzen-review/
They lock the CPU clock for some of the tests to test IPC only
 
  • Like
Reactions: fs123
like this
I'd really like to see a test where all cpus are clocked at 4ghz. It just feels weird to see a benchmark test where 1 cpu is clocked 1ghz higher and then claim it's faster in benchmarks than the one clocked at 4ghz. Second, I've shopped around for a new PC build and with the price difference between and AMD build and an Intel build, I can get a gtx 1060 with an Intel build but buy a GTX 1070 with my AMD build for just a few bucks more, depending on the card vendor.
This is an enthusiast website and we tested as such. Quite frankly, locking in clock on these would not be right since these do not run at the same speeds across cores in stock form.
 
I'd really like to see a test where all cpus are clocked at 4ghz. It just feels weird to see a benchmark test where 1 cpu is clocked 1ghz higher and then claim it's faster in benchmarks than the one clocked at 4ghz. Second, I've shopped around for a new PC build and with the price difference between and AMD build and an Intel build, I can get a gtx 1060 with an Intel build but buy a GTX 1070 with my AMD build for just a few bucks more, depending on the card vendor.

While were at it all Ryzen 7 CPUS should be tested with only 4 cores enabled in multithreaded tests against mainstream i7 processors.
 
Wow, thanks for the excellent review and all the work to capture all that data! Glad to have Ryzen's performance profile confirmed. Only surprise for me was how badly it did for mGPU.

This gives me a lot of hope that AMD will be able to get back into the CPU game and keep the pressure on Intel. If AMD can get a Ryzen follow-on with clocks up near 5 GHz and 3200+ DRAM things will be even more interesting. Seems like a reasonable goal for the next version on the same process, not a slam dunk but definitely a possibility. For now the price pressure is real, and has made Intel add a bunch of 6+ core chips to their consumer roadmap.
 
Thank you for the [H]ard work on the article. but i do not consider it the DEFINITE guide. Because there is a real world scenario where i am still at odds with the 7700k vs Ryzen issue:

- what happens when a gamer is multitasking in a AAA game, for example, live streaming the session with multi client audio?
 
This is why we love [H].

Thanks for doing such an exhaustive comparison.

Now if AMD can get that thing to 4.5ghz, Intel might finally get off their asses...
 
Thank you for the [H]ard work on the article. but i do not consider it the DEFINITE guide. Because there is a real world scenario where i am still at odds with the 7700k vs Ryzen issue:

- what happens when a gamer is multitasking in a AAA game, for example, live streaming the session with multi client audio?

Most of your serious streamers output to dedicated hardware and those who don't tend to use Intel QuickSync or Nvidia NVENC in OBS. I know that CPU encoding should have a bit better quality, but I don't think this would be noticeable if you were streaming it. Do you know of any streamers using Ryzen to stream using CPU encoding?
 
Thank you for the [H]ard work on the article. but i do not consider it the DEFINITE guide. Because there is a real world scenario where i am still at odds with the 7700k vs Ryzen issue:

- what happens when a gamer is multitasking in a AAA game, for example, live streaming the session with multi client audio?
That is why it is not named the Definitive Guide on Desktop Gaming and Streaming.
 
Most of your serious streamers output to dedicated hardware and those who don't tend to use Intel QuickSync or Nvidia NVENC in OBS. I know that CPU encoding should have a bit better quality, but I don't think this would be noticeable if you were streaming it. Do you know of any streamers using Ryzen to stream using CPU encoding?

I know nothing™, but this was a complain of 7700k owners on the ryzen thread here.
 
Appreciate the work that went in to this article. Lots of boring, repetitive data collection.

I would like to see how many fps are lost by dropping from the $300+ Intel i7 and AMD Ryzen 7 CPUs down to the $200ish Intel i5 and AMD Ryzen 5 CPUs. My prediction is that console ports won't see much loss at all, and the $200 CPU remains the sweet spot for mainstream gaming.
 
  • Like
Reactions: ecktt
like this
Great read, thanks for all the hard work that went into this.

The value side of me loves seeing the longevity of Sandy. Of course I'd like to be able to upgrade more frequently, but this article shows (from a gaming perspective) that I don't really need to. That and just from a personal usage standpoint my current system continues to meet my needs.

However, I agree with the conclusion that if I were to upgrade now, a Ryzen chip would almost certainly be the way to go since I ask my main rig to do so much more than gaming nowadays. Spend less on the CPU while still getting more cores vs Intel and feeding that savings into a better video card? Sounds like a win.
 
That's a really good review - thanks for that.

How long will it REALLY take for AMD to finish optimizing Ryzen? My question is ... why has it taken AMD so long to wait until now to finally decide to begin optimizing?

I do all the multi-tasking, content creation, video etc, ie "multiple cores with a box that is also your gaming rig" is me.

I'm in a catch-22 because I want the best gaming rig but, I also have to work too. I hoped that AMD's "Simultaneous Multi-Threading (SMT)" would've actually done something. It just seems like Intel is the better choice because a 370 mobo and a 1700X are not cheaper than an Intel z270 and a 7700k.
 
That's a serious shitload of benchmarking time. I appreciate the effort and the dual conclusions at the end seem to do a serious topic justice.
Thank you both.
 
Very nice roundup.

Only other thing I think would be interesting is to run these test while simultaneously encoding/live-streaming the content you're playing.
 
Great review guys, this is a real world comparison between the three CPU's and I am pleased to see that the Ryzen didn't disgrace itself, although I agree with your final comment about both companies not making enough progress over recent years.
It would have been interesting to see a slide with all CPU's clocked the same, say 4ghz, to see how much difference IPC made on its own with clock speed advantage removed.
 
Thanks Kyle.

What this really told me is that my 2600K that I purchased from Newegg back on 7/22/2011 was probably the best , longest lasting, computer purchase I have ever made. Sure I've upgraded the GPU a few times and am now on the 1080Ti but other than that it's the same system I built almost 6 years ago and it's still very competitive with the latest and greatest from both companies.

I've had the itch to upgrade for a while now, but there really isn't anything to truly upgrade too.
 
Very nice write up.

I must say, really surprised how close in many of those test Ryzen was, considering a deficit of 1 GHZ.

After reading this article makes me realize that my gaming PC will stay the same ([email protected]) until AMD can pass the 4GHZ barrier. Just because I dont use that system for nothing else but gaming, no streaming or other programs besides Defender.


as the article says we live in the now and not what might happen 3 years down the road

Agreed, but also remember, for a bit now, current consoles are 8 cores machines (granted, very weak cores), but the thread counts are there and could translate in a change in favor of CPUs with more cores.

tickles my curiosity is seeing an 7700k vs Ryzen on a "daily driver" computer that has various things running in the background.

Couple of reviews in other places did this and even though the raw numbers weren't a clean home run for Ryzen, everyone did say that the overall experience was way better with Ryzen, especially against i5's.

Intel is now offering 12, 16, and 18 core i9 processors. This would never have happened without Ryzen.

It already existed, but it had a Xeon label on it and extremely expensive price tags.
 
i9 != Xeon.

Also, you're not overclocking a Xeon using multipliers.
I know, small settings that they decide to expose/enable after, but by all intent and purposes, they existed already before Ryzen, just not as regular or enthusiast CPU's.
 
Nice article, BUT:
I have an identical Ryzen and Sandy Bridge system as the ones you tested, except I test with a Nvidia 1080 at 1080p.
The only difference is my ram is at 3466mhz C14 which is easy now with AGESA 1.0.0.6 on Crosshair Hero VI, and as
you know the infinity fabric runs at 1/2 ram speed, so running ram at 3200 or better yet 3466mhz makes a HUGE difference
then what you are running at. And I can tell you right now that my system is faster than all 3 of those systems with 3466mhz C14 ram
in most games. On top of that I can have 3 browsers, 3 chat apps, anti-virus, and 10 other apps running in the background
and STILL not lose any gaming speed. The 8 cores is very impressive and no one games on a perfectly clean install, so your
results are skewed. And by the way, did you see the new 14% improvement in Tomb Raider? That was 14% with 3200 ram and
software optimization, just think what it would be with 3466 ram. I can't stress enough how much more important ram speed is to
Ryzen then to an Intel system. A 4 core 7700 is the new dual core, very soon it will be overshadowed by 6 and 8 core cpus in every
new game. Nuff said.
 
Thanks for the review!

Which RAM kit was used on the GIGABYTE AORUS AX370-GAMING 5?


Regards,
Let me get with Brent_Justice on that. Quite frankly, we have gone through so many RAM kits in the last couple months, I truly do not remember. I had one kit on that board that did a solid 3200, but once the AGESA changed, I could never get that same, or any other kit I had at the time to do the 3200 again IIRC.
 
I don't think any of the games tested have been optimized for Ryzen. There are a few games such as Prey and ROTTR which have Ryzen support now and show very good performance on par with the best I7's. Maby those could be included in a future update.

Also the AGESA 1.0.0.6 bioses support faster memory clockspeeds as well as lower latency which will gives some more performance.

Finally, the test mentions the IPC of Ryzen being not as good as Intel but this issue can't seriously be determined when the compared cpu's have wildly different clockspeeds. The main reason Ryzen cannot match the Kabylake is due to clockspeed and in the future I expect newer steppings or process improvements could improve that alot.
 
I don't think any of the games tested have been optimized for Ryzen. There are a few games such as Prey and ROTTR which have Ryzen support now and show very good performance on par with the best I7's. Maby those could be included in a future update.
Then I guess AMD needs to step its game up and get some of the most popular recent games on its "optimzed" list. Prey will likely be added to our suite. It was not out when we started testing.

Also the AGESA 1.0.0.6 bioses support faster memory clockspeeds as well as lower latency which will gives some more performance.
I have been working with Ryzen systems for months. You can go back and read our experiences with the last Ryzen motherboard we did. Its AGESA is a big issue.

Finally, the issue of Ryzen IPC can't seriously be defined when the compared cpu's have wildly different clockspeeds.
Yeah, I can seriously define Ryzen IPC. I would suggest that I am in a very good place to be making statements about it.
https://www.hardocp.com/reviews/cpu_processors/1/amd
 
I don't think any of the games tested have been optimized for Ryzen. There are a few games such as Prey and ROTTR which have Ryzen support now and show very good performance on par with the best I7's. Maby those could be included in a future update.

Also the AGESA 1.0.0.6 bioses support faster memory clockspeeds as well as lower latency which will gives some more performance.

Finally, the test mentions the IPC of Ryzen being not as good as Intel but this issue can't seriously be determined when the compared cpu's have wildly different clockspeeds. The main reason Ryzen cannot match the Kabylake is due to clockspeed and in the future I expect newer steppings or process improvements could improve that alot.

I like Ryzen and think it's a great step for AMD, but it has lower IPC then Kabylake around 5% to 10% depending on the benchmark at the same speed. AMD has a good SMT and allows it to catch up against Intel the more threads that are involved. Also supporting higher memory speed and actually being able to hit them is something totally different as well and they are all beta bios at the moment. What Kyle showed you is what speed each chip is known to hit in the review and how they perform and what was learned was the GPU is still far more important then the cpu. The only thing I will say the future is brighter for Ryzen 1700 then the Intel 7700K as newer games are starting to make use of more cores.
 
Solid results! Just proves what i want to believe most of us were thinking. Games are still not CPU limited so, just upgrade the video card.
 
I'd be very interested too see how much speed AMD gets if DRAM overclocks of DDR4-4000 are possible. Now that the latest AGESA is out, we may finally know.

This is what you get in some games with Ryzen.

3593602e_17098594_1585065421522698_2471495432237228452_n-800x375.jpeg


I wonder how well this scales past DDR4 3200.


I agree with everyone else though - very interesting review.

AMD will be releasing a clockspeed and IPC bump with Ryzen 2 due in 2018, while we only expect the 14nm++ from Intel. The 10nm Intel CPUs by Intel's own metrics will be slower than the 14nm++; not until 10nm+ will they be faster.
 
Great review and thank you for doing it. Pretty much answers all the myths and rumors that have been floating around for the last several weeks with real data...nvidia driver performance, smoothness of gameplay, and actual game performance in a gaming environment.

I've been using an i7 3930k since early 2012. If an R5 1600 is simply two fewer cores than an R7 with the same IPC and o/c freq, it's exactly what I've been waiting for to step up to m.2 and USB 3.1 gen 2. Socket 2011-3 and Intel's price gouging is the #1 reason I've been milking this old SandyBridge cpu. The 3930k has been a great chip, but I had no reason to upgrade it.

**Edit: without going back to 4 cores

**Edit 2: removed PCIe x3
 
Last edited:
Amazing article. One that I think is going to reverberate around the pc gaming world.

My main takeaway is the amazing power and longevity of the sandybridge chips.

Wow. Every stone unturned. I feel dirty, like I got to see something I wasn't supposed to and I like it. Awesome review.

Intel should be ashamed. Seeing those 2 lines hug each other across that many generations of 'improvements' is shameful.

Hopefully Intel will send you a review pack with all those new i9's to evaluate and torture, although you mentioned you would probably have to source your own. I can wish though!!

Yeah. The test also showed how little 6 years of new cpu's have gotten the gaming community.....basically. Of course there are improvements but seeing how close the differences were between sandy and kaby, it almost didnt impress enough, near a letdown although I have already known that the performance was gonna be close. But really! That close? Guess next system will be based on a x99 or x199 platform., but since the improvements are based <effectively>on more cores and threads, just go with whatever feels right at that time.
Good article. If time and interest allows, you could reformat the numbers to clearly point out the differences between sandy and kaby for a second article.
 
I hear ya. All that [H]ard work for it all to become kinda mute with the new AGESA.
Don't believe everything you read on the internet. Remember we run REAL GAMING scenarios. Not a run it as fast as you can canned benchmark. There is a big difference there.
 
Back
Top