It would depend on what case you have and how much voltage you're going to be putting in. If you have an mATX case like me with the SG10 which is on the small side with unique cooling then it may something to consider to a degree but i don't think you need to worry too much as long as you have decent cooling for your overclock.
-
Hey, guest user. Hope you're enjoying NeoGAF! Have you considered registering for an account? Come join us and add your take to the daily discourse.
AMD Ryzen Thread: Affordable Core Act
- Thread starter ·feist·
- Start date
AM4 will last for 4 generations.
DemonCleaner
Member
thanks.
am leaning more to something like this atm: https://www.alternate.de/MS-TECH/Crow-Q1-green-Desktop-Gehäuse/html/product/1205454?
maybe without the excessive LED stuff
overclocking depends on how power hungry vega will be. on CPU overclocking: wanted to give the spire cooler a try. will eventually try a 3,9 GHz all core overclock, but just in case that won't affect the climate in the case too much. the system does not need to be whispering quiet, but i don't want the coolers to run around 100% either. guess the heat density in such a small case could become high rather fast??
Glad you found and resolved your problem in Bayonetta.
HPET usage is a contradicting mess on Ryzen so far.
- Until last version Ryzen Master enabled HPET on installation.
- But AMD itself still suggested with the preview samples: Make sure the system has Windows High Precision Event Timer (HPET) disabled. HPET increases the polling resolution of the system's timer for certain performance monitoring utilities and the increased poll rate can compromise everyday application performance.
- E.g. using HPET decreases the score in PC Mark from 4,147 to 3,966 (-4.4%)
- But without it changing REFCLK at runtime for OC introduces RTC bias that makes local benchmarks unreliable.
So it seems for reliable local benchmarking without HPET the OC should be applied in BIOS and not changed at runtime. It's odd that TSC is not used (not a sufficient replacement?) to ensure that the irregular behavior with and without HPET don't happen.
DemonCleaner
Member
is there a list of ryzen compatibel 3200+ RAM? heard one should go with samsung. not so sure how to identify which chips manufactures are actually using?
In the end RAM support is motherboard specific so your best bet is looking what RAM the manufacturers confirmed to be compatible with their respective boards.
Sectorseven
Member
I haven't seen any comparisons between the R5s and R7s, do the R5s perform better in gaming? Or are they basically the same aside from the core count?
or until DDR5 is released
Same, but less cores does not impact games a lot. 12 threads is more than enough.
R5 1600X and R7 1800X should be exactly the same aside the core count. For the other models the base and turbo frequencies differ.
so what exactly is going on with Ryzen? Every single review has incredibly conflicting results like Linustechtips and Paul's hardware having very different numbers with the only real change being the gpu. Some put it about on par and slightly above intel's current gen i5s (occasionally the i7 as well) while others completely cement it well below an i5 on all fronts for gaming. Are the Ryzen 5 cpus actually good?
Yes, it has been a mess.
It had been suggested that the difference between having HPET enabled or not was about 5% in benchmarks.
It seems like these huge performance drops may be an issue specific to Bayonetta, as it also happens on my 2500K system as well if I force HPET on, but dropping from 60 FPS to below 30 FPS is not exactly slight.
In other games it does seem to have been maybe 5-10% but it generally seems to be helping most games where GPU utilization dropped below 100%.
Firewatch did not improve much, but Dishonored 2 is a lot smoother now - so long as I restrict myself to 'high' quality textures and cap the game to 60 FPS.
It's not perfect but seems like it may actually be playable now.
The current version of Ryzen Master still tells me that I need to enable HPET so I don't think they have released the updated version yet.
Either that or the UEFI version for my system does not have the required AGESA update yet.
I don't think many people are going to be changing their overclocks after booting anyway - at least not people here that just want a system for work/games rather than overclocking being the hobby itself.
Supposedly all Corsair 3600 or higher kits are all Samsung B-Die.
I bought a 16GB 3600C18 kit which was the cheapest available here, and it's worked perfectly with my Crosshair VI Hero using the 32x multiplier or at 3600MT/s using BCLK overclocking.
It does not cold boot at that speed though. 2666MT/s or 2933MT/s have had no issues cold booting.
HPET may be negatively affecting performance in some of these tests if they ran Ryzen Master, which could explain the conflicting results.
So far, I'm pretty happy with my 1700X - it's amazing for non-gaming tasks, and pretty good for gaming too. (and better now)
But I've only been using a GTX 1070 and the performance may not hold up with a faster GPU.
The 6-core Ryzens seem like a pretty good deal compared to a 7600K to me.
I would find it very difficult to justify buying a 4c/4t CPU when you can get a faster 6c/12t CPU for less, and it doesn't seem to perform significantly worse in games with any of the tests I've seen, even if it's not always fastest.
Aside the low end 1400 the Ryzen 5 CPUs seem fine to me. Gaming benchmarks are not as consistent as other benchmarks as they add a lot more variability and interdependency, the GPU and its driver, game optimizations, hardware choices and OS settings (like above mentioned HPET) and the combination of them all can introduce unexpected bottlenecks that didn't happen before since Ryzen does a couple of things that haven't been done before in CPU design so software is not yet prepared for it.
Edit:
Just a random guess, maybe you can confirm it either way: I just recalled stock Dark Souls 1 always trying to lock frequencies to 60 fps when possible, then 30 fps etc. May Bayonetta also is not unlocked below 60 but drops straight to 30 as soon as fps deviate too low below 60 to stick to it? If so the relative small performance difference of 5% incurred by HPET may have been sufficient to cause that huge drop.
the 1600 and 1600x are a god send, if you ask me.
Whatever choice you make, always have in mind how long you keep your system and most importantly how games will be designed in that time period.
In my case, 3 years. I cant see the i5 performing like the 1600x 1 to 2 years down the road. At 3 years, lol
I was playing with G-Sync and also tested V-Sync off, so it wouldn't cause that to happen.
It was running at around 29-45 FPS most of the time, not just immediately dropping from 60 to 30 as soon as it dropped below 60, like double-buffered V-Sync would do.
Disabling HPET means that it never drops below 60 FPS during gameplay now.
Also, if you do have any habit of keeping other apps open while you game, a flash website or whatever, with an i5 you are losing performance today. Not 3 years down the road, but today.
Sectorseven
Member
Ignoring price, what about R5 1600X vs. R7 1700, or would that be entirely dependent on what the 1700 could overclock to?
That's really odd. Makes me think the added performance by disabling HPET helps staying at 60 fps, but the cause for the huge drops is something else.
Edit:
Aside core count the big different between those two is binning. With the 1600X 4GHz is part of its stock turbo even without overclocking. 1700 on the other hand is the bottom end of Ryzen 7, low power consumption at stock but overclocking beyond 3.7GHz will increase consumption exponentially and going beyond 4GHz is likely infeasible. 1600X is great for single thread performance and 1700 is great for core count.
It gets worse if you are using Windows 10's Game Mode.
Game Mode dedicates 3/4 of your CPU to the game process, and moves all system/other processes onto the remaining 1/4.
- With an i5-7600K, that means the game only has access to 3 cores/threads.
- With an i7-7700K the game has access to 3c/6t, with 1c/2t dedicated to the system.
- With an R5-1600X the game has access to 9 threads, 3 threads dedicated to the system.
- With an R7-1700X the game has access to 6c/12t, with 2c/4t dedicated to the system.
Of course you don't have to use Game Mode, but with the R7-1700X, even when CPU usage is generally low (only using a few cores and less than 10% overall CPU usage) and there's not much running in the background, I still see a small improvement in performance/consistency by enabling it in some games.
Of course, if a game can take full advantage of all 8c/16t, it is often better to leave Game Mode disabled.
I don't usually have enough background processes running to justify reserving a full 2c/4t.
I'd prefer if this could be configured so that I could limit Game Mode to using 1c/2t rather than 2c/4t.
I think it's actually a problem with the game, since it also happens on my i5-2500K system if I force HPET to be used.
But disabling HPET brought meaningful performance improvements to all games across the board - just not 30 FPS improvements.
5-10 FPS is the biggest difference I've seen so far, but there's been a big reduction in stuttering in some games - notably Dishonored 2.
Since the R5 CPUs are just lower-binned R7 CPUs with cores disabled, if money is not an issue and you plan on keeping the system for a while, I would pick the R7.
Well AMD was the one that said 4 generations, so hopefully they stick to their word. They would surely have a good indication on roadmaps for memory technologies when making such promises.
Bare in mind that AM4 could keep getting new CPU generations (in parallel), even if a new platform AM5 was to be released with DDR5 and other technologies.
Could be like the phenom 2, where the socket AM3 chips were backwards compatible with AM2+, as the phenom 2's memory controller supported both ddr2 and ddr3.
Don't the R5 full 16MB L3 memory for all 6 cores compared to the R7 8 cores? Won't that help in the future when games are more optimized for Ryzen?
I think you'll benefit more from having two additional cores than 0.67MB more cache per core.
You can also disable cores on an R7 if you really wanted to, though I'm not sure if that also disables cache. I would think not, since that's basically what they're doing with the R5.
That seems like it would ok for some overclocking probably not a lot especially with a non-blower style GPU. I would get a 120mm fan to put at the back for exhaust to complement the 200m at the front.thanks.
am leaning more to something like this atm: https://www.alternate.de/MS-TECH/Crow-Q1-green-Desktop-Gehäuse/html/product/1205454?
maybe without the excessive LED stuff
overclocking depends on how power hungry vega will be. on CPU overclocking: wanted to give the spire cooler a try. will eventually try a 3,9 GHz all core overclock, but just in case that won't affect the climate in the case too much. the system does not need to be whispering quiet, but i don't want the coolers to run around 100% either. guess the heat density in such a small case could become high rather fast??
Looks like GPU power cables could cause at issue with clearance depending on how thick they are.
Thanks for the info guys.
Thinking of upgrading my pc this year as I'm still rocking an i7 3770k and these results are just messing with my decision(s) lol
https://www.3dcenter.org/news/amd-r...tate-zur-anwendungs-performance-im-ueberblick
Every review listed below can be found here, along with *several* others: http://www.neogaf.com/forum/showpost.php?p=233774369&postcount=2444
Every review listed below can be found here, along with *several* others: http://www.neogaf.com/forum/showpost.php?p=233774369&postcount=2444
Black = Intel
Red = AMD
PdotMichael
Banned
With AMD not changing boards like underwear - going with the more budget solution is a reasonable strategy.
People shouldn't forget that.
People shouldn't forget that.
Level1Techs [YouTube] Gameplay/Frametimes of GTA V on Ryzen & Intel (3/3)
The Gigabyte AB350N-Gaming WiFi Mini-ITX was announced, but has not yet been shown:Is that the only ITX board that's been announced?
Im waiting for this so next month I can do a ITX build.
http://www.neogaf.com/forum/showpost.php?p=232090121&postcount=1815
This has been added to the OP, along with some of the other links that discuss HPET issues and why it should be disabled (including AMD's own "Tips for Building a Better AMD Ryzen System" community update).
thelastword
Banned
Here is the R5 1400 vs the I5 7500, which should be a bit stronger than the i5 7400, they trade blows, but notice that the memory on the R5 1400 is only 2400MHz DDR....Now if it had 3200MHz memory or 3600MHz memory, performance would be even better..., but it's performance here is quite solid...
Ryzen 5 Reviews Part 1: http://www.neogaf.com/forum/showpost.php?p=233774369&postcount=2444
Ryzen 5 Reviews Part 2:
LanOC Reviews Review: AMD Ryzen Stock Coolers [Wraith Spire and Wraith Max]
3DCenter AMD Ryzen 5 Launch Reviews: The test results for the application performance at a glance [German] | [URL="http://www.neogaf.com/forum/showpost.php?p=233887127&postcount=2526"]Translated Charts[/URL]
Digital Trends AMD Ryzen 5 1600X review: AMDs aggressively affordable Ryzen 5 1600X is the first six-core chip for everyone
Digital Trends AMD Ryzen 5 1600X review: AMDs Ryzen 5 1500X forgets to include the best feature
GameSpot AMD Ryzen 5 1600X and 1500X CPU Review: Power is in the eye of the beholder.
HardwareSfera.com Review: AMD Ryzen 5 1400 quad-core and eight-thread processing [Spanish]
MadBoxpc Review CPUs AMD Ryzen 5 1500X and Ryzen 5 1600X [Spanish]
Madshrimps AMD Ryzen 5 1600X Processor Review
Madshrimps AMD Ryzen 5 1500X Processor Review
Modders-Inc AMD R5 1600X 6-Core and R5 1500X 4-Core AM4 CPU Review
Neoseeker AMD Ryzen 5 [1600X & 1500X] Series CPU Review
PC Gamer AMD Ryzen 5 [1600X & 1500X] bringing competition to the mainstream - Fewer cores, a lower price, and firmware tuning turn Ryzen into a contender.
TechRadar AMD Ryzen 5 1600X review: High-powered CPU for a mid-range price
TrustedReviews AMD Ryzen 5 1600X and 1500X review
Ryzen 5 Reviews Part 2:
brain_stew
Member
4ghz 1600 overclock on stock cooling:
https://youtu.be/L4_4Tpa_cIQ
The 1600 is definitely the deal, it's a fantastic all around CPU for the money.
https://youtu.be/L4_4Tpa_cIQ
The 1600 is definitely the deal, it's a fantastic all around CPU for the money.
Can anyone clarify something to me [I don't have time to watch the video now], when Ryzen nonX CPUs are overclocked, are they then always run on those max clocks? Will they lower their clocks during calm periods?
TC McQueen
Member
If you have P State overclocking, yes. If not, I'm not sure.
Makes me wonder why Gamers Nexus recommended the 1600X over 1600.
Dictator93
Member
IIRC, GTA V has a bug after a certain framerate threshold where it stutters on 4C/4T CPUs like the i5. So that test and those minimum framerates are not very representative.
metareferential
Member
Maybe in the US (with 16 gb of ddr4 I doubt it anyway). Everywhere else in the world you wouldn't be saving that much.
thelastword
Banned
As a cost cutting measure, I think the 1600 is definitely a better buy...At $219, you get a CPU plus a wraith cooler...The 1600x comes with no cooler, but with the thought that if you're purchasing such a CPU, that you would most likely have a Noctua or liquid cooler to go along with it....
So yes, for enthusiasts, the 1600x makes a lot of sense, because the X chips are binned a bit better, having higher clock speeds and XFR, but not only that, they will be able to overclock better in the future too......So if you can afford it, I can see why some would opt for a 1600x.....
For some reason though, I think AMD has a lock on how much those chips are overclocked, you realize that no one has been able to bypass clcokspeeds above AMD's XFR speeds of 4.1GHZ ??? (Water and Air)......Perhaps AMD will loosen the strains a bit with the new microcode in May, but even then, the 1600X will overclock better than the 1600 when that happens....
dr_rus
Member
Well, yes, but 7700K will be considerably faster overall in gaming so that's not a fair comparison imo.
When choosing between 1600/X and some 6000/7000 series i5 I would go with Ryzen though. It has it's issues but the additional threads alone make it a lot more "future proof" - as in it will most likely perform better in future games than modern i5s.
Papacheeks
Banned
Will this get fixed? In an update Bios or update for Ryzen Master?
No chance for further overclocks and not a lock by AMD, 4.1GHz is the known feasible upper limit for overclocking of Samsung's 14LPP process (a process optimized for low power usage mobile parts, thus the name "Low Power Plus") that GF uses for the Ryzen chips. Beyond that point you are producing more heat than additional performance:
The processing has the ideal consumption/performance ratio until 3.3GHz, above that the increase is no longer linear and the ratio becomes continually worse.
Well my situation is i do a bit of gaming, a bit of game capturing, a bit of video editing and encoding. I currently have a Haswell i7 4770 and am thinking of getting the Ryzen 1600 as it would seem to best fit my situation, yes i maybe losing out a bit on games but in real world term are you going to notice a difference between a game running 89fps on an i7 or 74fps on a 1600?, but i should see massive gains in capturing, editing and encoding!?
dr_rus
Member
Capturing of video stream is mostly videocard memory bandwidth limited AFAIK, encoding when done correctly on GPU's encoder doesn't cost anything on modern systems. For the editing though 6 cores can be faster than 4 but I don't know about "massive" when comparing 4770 with 1600. Haswell isn't that much worse than Kaby Lake in IPC and frequencies.
Depending on your task you don't want to rely on the poor quality produced by GPU encoders. If you're producing professional or semi professional video you're almost certainly encoding on CPU. The fixed function encoders in gpus are at their very best called "just OK".
Yeah, Linustechtips did a respective test, was posted two pages ago: https://www.youtube.com/watch?v=jludqTnPpnU
I'm not even talking about streaming. But, yes streamers often use a streaming box that uses a cpu encoder due to the better Quality/Bitrate ratio.
metareferential
Member
I doubt you'll see a significant upgrade in gaming performance. Multitasking surely would fare better.
Old but popular (overclocked) processors against ryzen new ones is something I'd like to see more frequently addressed.
The video does a pretty good job showcasing the (lack of) quality produced by GPU encoders.
thebigmachine
Member
At the same bitrate, compensate with a higher bitrate. Encode so fast it will make your head spin vs cpu.