安裝 Steam
登入
|
語言
簡體中文
日本語（日文）
한국어（韓文）
ไทย（泰文）
Български（保加利亞文）
Čeština（捷克文）
Dansk（丹麥文）
Deutsch（德文）
English（英文）
Español - España（西班牙文 - 西班牙）
Español - Latinoamérica（西班牙文 - 拉丁美洲）
Ελληνικά（希臘文）
Français（法文）
Italiano（義大利文）
Bahasa Indonesia（印尼語）
Magyar（匈牙利文）
Nederlands（荷蘭文）
Norsk（挪威文）
Polski（波蘭文）
Português（葡萄牙文 - 葡萄牙）
Português - Brasil（葡萄牙文 - 巴西）
Română（羅馬尼亞文）
Русский（俄文）
Suomi（芬蘭文）
Svenska（瑞典文）
Türkçe（土耳其文）
tiếng Việt（越南文）
Українська（烏克蘭文）
回報翻譯問題
討論區規矩與守則
檢舉此篇文章
Games start at 11:39 ^ again not a huge diff you'd get more umff in your GPU dollar.
100%, was looking into this just yesterday and its exactly what i am doing. Going with the gskill ripjaws v series 3600.
It's only going to matter in non-gaming situations.
As Autumn said, it doesn't make a meaningful difference. You're paying more for 3 FPS when a cheap 3200 MHz CL16 kit (i.e. Corsair Vengeance LPX or G.Skill Ripjaws V) is more than satisfactory. I've overclocked RAM and tested 3600 kits, the difference in gaming is next to nothing.
3600 MHz only makes a discernible difference in worthless synthetic benchmarks and specific workloads that depend heavily on CPU+RAM, like work related programs that many Ryzen 9 and Threadripper users would be running.
AMD only recommends 3600 MHz because they probably have a deal with G.Skill, because Trident Z Neo was manufactured and touted as "the" kit for Ryzen 3000. They did the same thing with Flare X, but the difference here is the memory compatibility is worse than random 3200 MHz kits.
Lastly, and quite frankly, Hawkens doesn't know Ryzen like I do. Having used Zen, Zen+, and Zen2 CPUs quite extensively with multiple motherboards and RAM kits, I can safely say that the 3600 MHz recommendation is ♥♥♥♥♥♥♥♥. He can shake his head at people recommending only 3200 MHz, but at the end of the day, he's not the one building the system and can't guarantee that 3600 MHz kits will work with the advertised specification; you'll most often have to crank up DRAM voltage and other voltages and/or reduce clocks/increase timings to get it to work. Games can easily crash more because of unstable memory, I've tested it myself and have confirmed it.
TL;DR -- 3600 CL16 is ONLY recommended if you're running R9 or TR4 and are using the CPU for heavier workloads that benefit from such a RAM frequency. Gaming is not nearly enough of a workload that RAM above 3000 MHz isn't enough.
If people are interested they can Google for ryzen dram calculator, but you must know what you're doing before tweaking otherwise you can cause all kind of problems to your pc, I can't stress that enough...
https://youtu.be/iH3qq_mSxTM
Also the case isnt to much of glass either so,
You say this...
And yet LTT says...
I have the MSI MEG Unify x570 running with 3600 CL16 Neo flawlessly since day one...
I have a budy who built nearly the same system (Unify/3900x) but with the GSkill Tident non-Neo non-RGB basic 3600/CL16 kit, Also has no issues what so ever...
So, I have my own exiperiance with no issues, I have my buddies experiance with no issues, and I have LTT telling me thats what I want if I want no issues...
And I have you telling me it causes issues...
Higher than 3200 was spotty on x370/x470 but it is not spotty on x570 with ryzen 3000... On that platform it is somtehing "that you *know* will work."
As to performance, proper ram with tight timings on Ryzen often buys 5-15% in most titles, specially with manurally tuned timings. Even running stock a good kit vs a mediocre kit can pull 5-10% or greater in performance increase.
Considering Intel fanbois regularly agrue intel is worth it for gaming over intels 3-5% lead right now, its more than reasonable to make the same argument that the RAM is worth it for 5-10%, expecially in markets like the US where the cost difference is meaningless.
1. That's only two instances where it worked, and there's no guarantee that it will always work.
2. It's unnecessary to spend more on 3600 CL16 when 3200 CL16 is FINE. There's not enough of a benefit for gamers for it to be worth it, and if you actually watched FPS comparisons between 3200 and 3600, you'd see that.
3. It depends on what X570 motherboard you buy, cheaper ones are more likely to suffer.
4. It doesn't increase performance by nearly that much. Literally, just save a bit more on your board and RAM, and you can afford a better GPU.
Watch the videos and you'll see how pointless it is. 20$ USD for RAM and however much you'd spent for an X570 can add up and interfere with buying a better GPU, which will do MUCH more than wasting money on a 3600 kit.
https://www.youtube.com/watch?v=kP9F0h7qP_g
https://www.youtube.com/watch?v=S0WZNwb_6XU
5. You don't seem to realise that many people can't afford to spend as much as you did. Most people can only fit, at most, a Ryzen 5 3600 with 16GB DDR4-3200 on a B450 motherboard with their budget.
2) No, its no neccisary, but then again, its not neccisary to leave multi-core performance on the table for the 3-5% (or les) Intel can offer right nwo in gaming due to IPC, and yet people make that argument (and purchase choice) all the time. Simple *fact* is that the small 5-10% improvment on Ryzen 3000 with proper RAM brings it into parrity with Intel on gaming, and *that* is just as much worth it as buying Intel is. Not needed, but def nice to have..
3) This is true, but that is always true. Dont cheap out on the motherboard. Kinda like cheaping out on the PSU, jsut *dont* do it. Period.
4) It highly depends on the title, but yeh, it *can* provide plenty of performance increase....
You are just cherry picking horrid examples... FFS, the second vid you link is comparing 3200 CL16 vs 3600 CL17... No fricken duh there is no change there, the 3600 is lattency gimped ffs...
The first vid is a *bit* more realistic, its showing 3600 cl16 vs 3200 cl15 (not likely to actually be purchased). Here you *do* see the fully expected increase in FPS not shown in the useless second video, and you can see that in most titles the better ram pulls 5+FPS or better. But again this video does *not* show a valid real world comparison...
Why? Because 3200 cl14 is just as much $$ (and nearly identical in performance due to timing) as 3600 CL16. They perform roughly the same, and cost roughly the same, because you are trading raw speed for lattency and get the same rough real world performance. Thus comparing those two is a non-comparison and non-realistic to what people buy.
Lookup Ryzen 3000 comparisons using *specifically* 3200 CL16 (cheap) vs 3600 CL 16 (pricey). *NOW* you will see the impact exactly as I have stated it.
Arguing 3200 is just as good, then linking back to vids that are comparing tight 3200 vs average 3600, or comparing average 3200 against loose 3600 is just wrong... You are playing an apples to orages game and trying to call them the same...
5) I would never argue for 3600 cl16 on a b450/r5-3600 build. Thats a budget build and should go with budget ram. OP is asking about a 3700x on x570 build. If he is already dumping on that then getting proper RAM is a no brianer. He has the extra 20 somewhere, even if its in high couch cushions.
I think this is the kit my buddy used on his unify/3900 build, black with no RGB right, but same timings as the Neo cl16 kit, about 178 for a 2x16 kit? Good ram!!
I'm comparing 3200 CL16 and 3600 CL16, not 3200 CL14.
3600 CL16 doesn't help as much as you think it does. Having tested the difference, I know this to be fact, and you seem to be under the impression that everyone is buying 3800Xs and X570s, when that is FAR from true.
The majority of Ryzen 3000 buyers are going for a 3600 or 3700X on a B450 Tomahawk MAX because that's what is most often recommended. They don't care about maximising FCLK, nor does it have enough of a benefit to gaming that it's even worth worrying about, when most games are only going to gain a few FPS at best, especially at 1440p.
But you are not comparing 3600cl16 vs 3200cl16, neither of the vids you linked to did...
CL16@3600 does make a big difference when you compare it to CL16@3200. Only when you look at 3200 with CL14 can the 3200 kits match the 36000, or if you look at a 3200 cl16 vs an even more lose 3600 kit, and in both the reviews you linked to that is what happened.
You are right that allot of Ryzen buyers are going r% budget builds, and for them 3200 is the best match, but for anyone going 3700x or greater *and* investing in x570, its just stupid to opt for anything less than 3600cl16.
And since we are talking about the OP, and they *ARE* going x570/3700x, then it would be pretty silly not to match the rest of the system up to spec.
If you're only using XMP, the difference is not gonna worth it, you have to manually tweak your hardware to gain anything significant, and this is not what average user can do.
https://www.youtube.com/watch?v=i0CZ1GJW24c In more demanding titles like AC:O, it makes no difference. All kits in the video are CL16. Only in the first title is there a real significant difference because it's easy to run, and it's 720p.
Also: Higher DRAM frequency still does more than lower latency. Even if the 3600 kit has slower access times, it'll still win because it's able to do more during that slower access time. You're only going to see a really bad difference between 3200 CL14 and 3600 CL18
Entire argument and video is Intel making it 110% invalid to this entire dicsussion.
Do you *not* know about the importance of memory speed and timing on Ryzen?...
Here:
https://www.youtube.com/watch?v=7w1EGPZUESU
Enjoy this OC 3rd Gen Ryzen guide by linus media group. Notive how nearly *half* the video focusses entirely on RAM, timing and its impacts on CPU performance, and how the sweet spot for speed *is* 3600. Then notice how with an FCLK to match a 3800 kit they were able to pull an aditioal +13fps average and +19fps 01% minimums by doing *nothing* but tightwning timings...
Dont even bother bringing an intel only comparison video (specially from 8th gen or newer) when it is widely known Intel gains little if any (specially back then) for RAM when AMD has had cpu performance tied to RAM due to infitinity fabric for generations.
1. I specifically remember you asking about Ryzen a few months ago, and I literally gave you the answers. Now you're preaching like you're an expert on Ryzen because you have a 3900X because of ONE experience. I've worked with the 1700X, 2700X, and 3900X on different motherboards and with different RAM kits with each CPU. You haven't. I know for a fact that you're reading into it more than necessary.
2. Actually, it's not the DRAM frequency and timings that matter, it's the infinity fabric clock, or FCLK that actually matters. It can be tweaked on slower RAM to work faster than it would when you just leave it alone. You can make 2133 MHz RAM run decently well compared to 3000.
3. Do you not realise that I've TESTED 3200 CL16 vs 3200 CL14 vs 3600 CL16 myself? Unless you're actually focusing on every little FPS you get from overclocking RAM freq+timings, you're not even going to notice a damn thing unless you're pointlessly obsessed with pointless benchmarks.
4. RAM frequency and timings are literally just as important for Intel as they are for AMD. They get the same bonuses from higher frequency and lower timed RAM as AMD Ryzen does. It's only slightly less pronounced because the IMC in Intel CPUs is arguably better and doesn't depend as much on RAM.
5. 1% and 0.1% minimums are not even important. The focus should be on 100% minimum ~ max.
EDIT: Also forgot that I use a laptop with a 2200U and 2400 MHz RAM.