-
Competitor rules
Please remember that any mention of competitors, hinting at competitors or offering to provide details of competitors will result in an account suspension. The full rules can be found under the 'Terms and Rules' link in the bottom right corner of your screen. Just don't mention competitors in any way, shape or form and you'll be OK.
Ryzen and Gaming results.
- Thread starter geordiejc
- Start date
The Tech Talk streams he does (sadly they are ending soon) with Jerry "Barnacules" Berg are awesome, tech news/perspectives with humour, probably the most comical moment was when they were discussing something about Microsoft (where Jerry worked as a senior software developer for 15 years) and they suddenly realised that Jerry was actually the lead developer on the automation software that later laid off him and his department XD
Unless they had a Gigabyte, Motherboard, where those reviewers got better overall results, better frametimes/minimums, and better memory frequencies. Even at 720p and 1080p.
There's a lot of variance between reviewers, especially when motherboard and BIOSs are different.
Why in their 7700K reviews do they ignore the x99 systems suddenly?
http://www.gamersnexus.net/hwreviews/2744-intel-i7-7700k-review-and-benchmark/page-6
Especially when focusing on highend workloads.
http://www.gamersnexus.net/hwreviews/2744-intel-i7-7700k-review-and-benchmark/page-5
I find it all so odd going through it. As mentioned before some reviewers even compared the 6900K vs 7700K at 1440p and 4K only. Forgoing 720p and 1080p results; and honestly Ryzen had more 720p tests than I've ever seen on x99.
Even when talking specifically about Watch Dogs 2 and CPU optimization they ignore the x99 CPUs suddenly. despite the game gobbling up more CPU cores.
http://www.gamersnexus.net/game-bench/2785-watch-dogs-2-cpu-settings-optimization-guide
Shown by ComputerBase
https://www.computerbase.de/2017-02/cpu-skalierung-kerne-spiele-test/#diagramm-watch-dogs-2-fps
Ole Bitwit Sums it up rather well. If the 6900K was compared like R7 CPUs, it would probably also earn the Gamer Nexus phrase of "An i5 in Gaming, i7 in Production"; except you know the 5960X and 6900K cost 50-70% more than Ryzen.
33:53 in it
https://youtu.be/3NJN-C1R9gA?t=33m53s
Last edited:
ive been wondering if some of the reviews were not so good because of motherboards, reviewers most impressed look to have used gigabyte
Good to see the improvements of the AMD 8350 4.0GHz turbo to 4.2GHz vs the Intel 2500K @ 3.2GHz that will turbo to 3.7GHz. I guess the majority of casual PC gamers don't overclock, so those results make good sense.
I would be interested in seeing the results of both CPU's at a similar overclock say 4.5GHz, this is Overclockers UK, not Stockclockers UK afterall lol.
Likewise I really want it to run perfect with at least 3200mhz RAM, considering it's only dual channel it needs to have fairly quick RAM imo. I'm wondering if Zen+ might be worth waiting for but I'd like to get on the ship when it's new and fresh.
slashing prices for no reason ? they just being competitive in a competitive time.its standard.why are they worried ?
they have the fastest processors.they have the market share.they dominate.they have new more powerul ranges soon.i bet they crying in there billions
they have the fastest processors.they have the market share.they dominate.they have new more powerul ranges soon.i bet they crying in there billions
It is mighty peculiar though, what with X99 and BWE gaming performance in respect to Skylake/Kaby not exposed to anywhere near the same scrutiny and criticism upon their release, and now suddenly Ryzen is getting slammed for not matching the 7700k and 720p gaming of all things! Of course, one could point towards AMD themselves making certain claims and putting it in on the 'gamer radar' in a way that X99/BWE never did, but I see that as grossly unfair. In my view, all that recent talk about Intel trying to influence reviewers and impact Ryzen sales with negative spin seems to have to come to fruition. Some may shout tin foil hat conspiracy, but the double standards here are quite plain to see. Wouldn't surprise me if pressure was applied and/or money exhanged under the table in order for
The thing I find hilarious is, yes the 7700K is better than Ryzen if you only care about gaming and nothing else, but in that case it's better than X99 too and X99 isn't being slated. Add to that, if that's all you care about then the i5-7600K will be just as good as the i7-7700K while costing less so you're only buying the i7 for epeen anyway.
In short, anyone looking at an i7 who doesn't need >2 GPUs should be looking at Ryzen instead right now unless they only want it for bragging rights, and this is something the AMD haters cannot reconcile so they try to find whatever angle they can to hate on Ryzen.
Quite, yet that doesn't seem to stop people digging in to it lol!
Agreed, but obviously this doesn't even speak to the motherboard/BIOS and optimisation issues that we know are holding it back slightly. Obviously it's unfortunate things weren't smoothed out before release, but it's hardly a case where Ryzen performance today is the same as Ryzen performance 2 months from now. The same can't be said of Intel. It's also an often sidestepped fact that future games will inevitably start to make further use of those extra cores which Skylake/Kaby lack. The mere fact Intel are going 6C with Coffee Lake suggests they see the writing on the wall there, yet all the Intel shills and fanboys are still arguing that 4C is all you need lol! Hilarious. Today perhaps, but looking ahead, no.
For some applications sure - these days games don't tend to have a lot of CPU specific programming and only 1-2 engine coders really go to that level of trouble, one of which is currently not even working in the engine coding business.
If you compare the source code for more recent game engines to older ones there is a lot less where they are utilising inline assembly to exploit specific capabilities of CPUs (partly due to the complications in respect to that and x64 processors compared to x86) and in the case of game engine performance compiler optimisations as a gross generalisation don't have the potential to significantly improve how well a game runs on Ryzen.
I think I have found where Ryzen's inconsistent performance comes from, it's based around the fact that Ryzen is two quad core modules and their cache configuration (specifically L3).
If you notice Ryzen has 2x8MB chunks of L3 cache rather than a single 16MB chunk shared between all cores like Intel (actually 20MB), I suspect this is where the inconsistent performance is coming from. The architecture is based on 2 separate quad core units (CCX) tacked together, each CCX has 8MB L3 cache.
TechPowerUp did some testing:
The latency going through the roof is probably due to paging to main memory when the 8MB L3 is filled.
So in situations where less than 4 cores are being utilised the cores only have access to 8MB L3, unlike the 6900K where however many cores are being utilised they have access to the full 20MB L3. I suspect this is also where the Windows scheduler could help, if a 4-core workload is spread across both CCX (2 cores per CCX) then you'll have a situation where the 2 cores (per CCX) have access to 8MB L3 between them, as opposed to 4 cores on a single CCX only having access to 8MB L3 and the other 8MB L3 pool going unused.
Apparently there is an interconnect between each CCX but it's slower than just paging to main memory:
One of the problems I see with this configuration is that in situations where you have 5-8 cores being utilised on the same operations data is going to be duplicated in both L3 pools which could theoretically halve L3 size in a worst case scenario, or worse still data will be transferred between CCX which is over 8 times slower than L3 itself. A good way to think of it is it's like Crossfire/SLI where two 4GB cards is still only 4GB rather than 8GB because each GPU (or in this case CCX) needs access to the same data, although in Ryzen's case if both CCX are working on different operations they won't necessarily need the exact same data present in L3 so effective L3 is going to vary between 8-16MB depending on how much data is duplicated. I'm not even sure what the point of the CCX interconnect is when it's faster to just page to main memory.
I think with this L3 cache configuration you're never going to get truly consistent performance but I can see how an optimised Windows scheduler could optimise CCX use better.
If you notice Ryzen has 2x8MB chunks of L3 cache rather than a single 16MB chunk shared between all cores like Intel (actually 20MB), I suspect this is where the inconsistent performance is coming from. The architecture is based on 2 separate quad core units (CCX) tacked together, each CCX has 8MB L3 cache.
TechPowerUp did some testing:
The latency going through the roof is probably due to paging to main memory when the 8MB L3 is filled.
So in situations where less than 4 cores are being utilised the cores only have access to 8MB L3, unlike the 6900K where however many cores are being utilised they have access to the full 20MB L3. I suspect this is also where the Windows scheduler could help, if a 4-core workload is spread across both CCX (2 cores per CCX) then you'll have a situation where the 2 cores (per CCX) have access to 8MB L3 between them, as opposed to 4 cores on a single CCX only having access to 8MB L3 and the other 8MB L3 pool going unused.
Apparently there is an interconnect between each CCX but it's slower than just paging to main memory:
One of the problems I see with this configuration is that in situations where you have 5-8 cores being utilised on the same operations data is going to be duplicated in both L3 pools which could theoretically halve L3 size in a worst case scenario, or worse still data will be transferred between CCX which is over 8 times slower than L3 itself. A good way to think of it is it's like Crossfire/SLI where two 4GB cards is still only 4GB rather than 8GB because each GPU (or in this case CCX) needs access to the same data, although in Ryzen's case if both CCX are working on different operations they won't necessarily need the exact same data present in L3 so effective L3 is going to vary between 8-16MB depending on how much data is duplicated. I'm not even sure what the point of the CCX interconnect is when it's faster to just page to main memory.
I think with this L3 cache configuration you're never going to get truly consistent performance but I can see how an optimised Windows scheduler could optimise CCX use better.
Last edited:
x99 doesn't get as much criticism because intel offer a separate platform which you can chose if x99 isn't right for your use case. AMD have gone for a single architecture which means it is going to be compared to both x99 and z270 for all use cases. It is better then x99 for most imo as it is such good value for basically the same performance (outside of fringe cases where you actually need all of the PCI lanes that x99 supports). Compared to x270 though, it falls a bit short which is disappointing for people that do only want a gaming PC. I think its safe to say that most of the flak that Ryzen is getting is from people who are letting their emotions get the better of them as they want to buy it but logically it doesn't make sense for them so they are stuck with intel. I'd love to switch to AMD myself but probably won't be for the foreseeable future.
x99 doesn't get as much criticism because intel offer a separate platform which you can chose if x99 isn't right for your use case. AMD have gone for a single architecture which means it is going to be compared to both x99 and z270 for all use cases. It is better then x99 for most imo as it is such good value for basically the same performance (outside of fringe cases where you actually need all of the PCI lanes that x99 supports). Compared to x270 though, it falls a bit short which is disappointing for people that do only want a gaming PC. I think its safe to say that most of the flak that Ryzen is getting is from people who are letting their emotions get the better of them as they want to buy it but logically it doesn't make sense for them so they are stuck with intel. I'd love to switch to AMD myself but probably won't be for the foreseeable future.
This is pretty accurate and I'm probably in this category apart from photo editing, which is minor compared to video. It just doesn't quite have enough to justify a switch from an older clocked i7 for me, although I'm still tempted. Should I not get along with it I'd keep the ddr4 and simply sell the CPU and board. The rest of the system could be turned to x299. My windows is retail so that's not even an issue with keys. Still tempted...
I'll probably end up waiting for skylake X and find out the 6 cores are £450 with decent boards starting at £275+. Or in other words, over £160 more than I'd spend for ryzen, months later, for a few more frames in the short term and two less cores so it's likely worse at anything outside of gaming
Last edited:
One of my friends works for certain well known tech giant and currently they have a huge team working on quantum computers ( had for last few years) - they actually have working product already but not for mass market... in a 2 or 3 years though all of our current tech could be obsolete anyway.
On a side note, I haven't been following am4 socket much but wouldn't 1700+decent mobo be cheaper than 7700k equivalent?