Ryzen 7 3800X reviewed:
https://www.tomshardware.com/reviews/amd-ryzen-7-3800x-review,6226.html
https://www.tomshardware.com/reviews/amd-ryzen-7-3800x-review,6226.html
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.
Where are all the 3800x reviews...they are in stock now so reviewers must have them....hmmm
Is there a similar comparison available for 2700x v 9900k?
The lower TDP is starting to appeal to me, although I'm not sure that the performance increase is worth it for gaming.
Cheers Ctf, my 3900X is in the post so I'll be able to see exactly how fast it is for my exact workflow. Though these results are both for mass exports of multiple photos at the same time which will generally favour multi core CPU's. I rarely if ever do anything like that so I can not extrapolate those benchmarks to my usage.
Nah, it was simpler than that, they just bought a Z390 board instead of re-soldering a new VRM/power delivery layout on to the boards
Those i9's getting beaten by a Ryzen 5 3600 is something.
A lot of reviewers did this and i don't think it's fair, i've been watching a lot of 3600 reviews and on the stock cooler its only boosting between 3.9Ghz and 4Ghz, my own on a 120mm AIO is boosting between 4.1 and 4.2Ghz, that's 5%.
To me something is off with that, Ryzen comes with a Box cooler, fine, use it, but then if you're going to use a £20 cooler for Ryzen why would you use a £150 cooler for the 9700K / 9900K knowing it wouldn't boost to 4.7Ghz / 5Ghz with anything less? a lot of them are also not clocking the RAM up on Ryzen.
Maybe AMD should stop providing Box Coolers to force reviewers to give these Ryzen chips a fair showing, eh?
Anyway, 3600 vs 9600K, i can tell you he has not got the 3600 running on the box cooler in this.
DF said:And then there's the overall package itself. The Ryzen 7 3700X is more power efficient than the Core i7 9700K and it doesn't require extreme cooling to offer optimal performance. In fact, the supplied Wraith Prism cooler is effectively overkill for the thermal output of the chip, and will contain the extra heat generated by overclocking (though this is limited somewhat as you won't get more than a couple of hundred megahertz extra out of the chip). In contrast, the Core i7 9700K doesn't ship with a cooler - but its overclocking headroom is a bit more significant, though not game-changingly so.
Fast memory of at least 3000MHz is recommended, and it's swiftly becoming a standard in the marketplace, and we're lucky in that AMD allows for overclockable memory to run on both high-end and mid-range motherboards, while at the same time, the use of the AM4 socket means that the 3700X should run just fine in the vast majority of existing boards out there. On top of that, the inclusion of PCI Express 4.0 support on boards using the X570 chipset means that future graphics cards and - more importantly - faster storage are now viable on a mainstream platform. PCIe 4.0 won't have a dramatic impact on gaming, but the fact that Intel's mid-range boards don't allow users to run their RAM at speeds beyond the chip specification is something that really has to change. If we can find a game that loses seven per cent of performance on an i7 9700K dropping from 3600MHz to 3000MHz, what would that drop be on a non-Z board where we're limited to 2666MHz memory bandwidth?
And finally, we need to talk about price. The Ryzen 7 3700X is £320/$330 up against the £379/$409 Core i7 9700K. Prices on the Intel chip are dropping in the US, but in the UK the price deficit as things stand pays goes a long way towards buying a 2x8GB 3200MHz DDR4 memory kit - and remember, all the cooling you'll need is already in the box and you won't have to pay over the odds for motherboard to run that memory at full frequency. Intel is faster in games (sometimes appreciably so, often not by that much) and it can overclock to 5.0GHz - the question is whether those advantages are worth what is - in real terms, system-wide - a big price premium.
I ordered one of these as an upgrade to my Ryzen 5 1600... it's great I can use the same motherboard and get a 30-60% jump in performance at the same 65 watts and don't have to spend more money on a cooler. Really nice showing for AMD.
Yeah, Hardware Unboxed in a separate review compared power consumption and clock speed AIO vs Box cooler, on the AIO Ryzen used a little more power, and boosted higher, what a surprise that isn't to me.
Ryzen 3000 behaves a lot like Pascal GPU's, the cooler you can keep them the higher they clock themselves.
Look at those reviews i posted, 3600 under a proper cooler. 3600 @ <4.2Ghz 498 FPS, 9600K @ 4.3Ghz 423 FPS that's a difference of 18% to the 3600, side by side video runs don't lie like slides so easily can.
5% or so might not sound like a lot but where they are this close it maters. https://www.techspot.com/review/1877-core-i9-9900k-vs-ryzen-9-3900x/
"Its not performing as well as it could with a proper cooler, but the point we want to make in this CPU performance review is its great value, AMD again with the lack of performance but hey ho at least they are cheap"
No, i'm sure AMD are sick of being seem as the "budget option" FFS let it stretch it's legs and put a proper cooler on it instead of strangling it and perpetuating the "budget plebeian option" that has plagued AMD for a decade.
CAT my board doesn't have PBO, it runs at <4.2Ghz without it under a 4 year old KRAKEN X31 at <65c. using PBO in Ryzen Master makes 0 difference.
I don't care about it being the value option, it's just plain a fast CPU, let it fly.
When you can pick up PC3200 ram so cheap, I doubt the expensive ram would even show a 5% increase across a broad range of games. Probably more like 1-2% max. So yeah, pointless spend.
MSI LiveStream Conclusion
- B450 Tomahawk will stay on GSI Lite and will not recieve the Click Bios 5 again, at least for Ryzen 3000 Series and upcoming. Users of 1000/2000 Series should stay on their BIOS.
- GSI Lite BIOS will not going to have OC profiles again. Update: They are looking into it. No promise. Quote: " ***MSI Gaming:*** just checking some bios release note, OC Profiles might be back in future GSE-Lite bioses "
- MSI is *now* aware of the problems regarding the Tomahawk and CPU Debug light issues and will investigate that problem. They hopefully have some new Infos next week, but no promise.
- "Old" B450 MB (including Tomahawk) will have Ryzen support until 2020 (?)
- If you just bought any B450 Board, you should return it and buy a MAX board instead. Its more "futureproof" for upcoming BIOS updates and its no hassle with Ryzen 3000 Series. (Official statement on livestream from MSI, wow.) Timestamp on stream: 1:40:29, you can watch it here -> https://www.youtube.com/watch?v=C_elcRHeVjI
- After i asked this question: "Will there be an option to RMA an not working B450 Tomahawk and recieve an B450 Tomahawk MAX (maybe with additional charge) ?" They closed the stream. Quote: " ***MSI Gaming:*** Sorry it seems the stream dropped, anyway we are out fixing your Tomahawk issues. thanks for joining this was the last topic anyway. Thanks for joining and see you next week, hope to have an update on Tomahawk... no promise."
The problem is they dont test everything and when they do its often flawed methodology, after getting some results back from a friend I can tell you intel outperforms in a few games we tested that are not twitch shooters. Now is the performance huge or at least big enough to be an issue, thats not necessarily the case, but reviewers games testing is extremely narrow, and they also tend to do only minimal testing on stuff that isnt content creation or gaming.
Of course what is significant and not significant is also a matter of opinion, the reviewers only state an opinion in that regard. Some people will spend insane money to gain a few %.
And thats why Intel is doing this 9900ks. Its more like 3000series in a way aka runs at MAX out of box.
cat the fifth just noticed your reply.
I am surprised you know many people with K chips that run at stock speeds, as whilst thats a valid way to run them, its a poor value for money deal as there is a premium on those chips price wise.
In terms of "safe" overclocks I think your point is fair, I know first hand there is silicon lottery losers, and with reviewers seemingly getting a decent chance of a golden sample, that sways things even further so e.g. there is people claiming every coffee lake can hit 5ghz, and if someone posts they only get say 4.8ghz using 1.35vcore then the end user must be doing something wrong it cannot be simply that they got a silicon loser. So for this reason I think using a 8600k or 8700k at 5ghz as a comparison isnt quite right either, its definitely vague, a safe overclock is probably what MCE configures so probably 4.7 for an 8700k and 4.3 for an 8600k. I have never heard of an 8600k not able to reach 4.7 but there is some pretty nasty 8600k's out there, one guy needed 1.38v to hit 4.7. I also suspect a fair few of these claimed stable 5ghz chips are not proper stable. My own 8600k will boot into windows at 5ghz, I can run aida bench, cinebench bench, play games. Browse the net etc. For a few hours. It probably would crash doing these activities eventually but I suspect some people will tolerate it providing its not too frequent. It of course crashed on stress testing.
If I was reviewing these parts I would likely test all 3 scenarios, shipping clocks, MCE clocks and "aggressive" clocks. 5ghz is aggressive on intel, even if you can get it stable they need very good cooling at those speeds. I would add a note MCE is whats realistically achievable, anything on top of that a bonus.
AMD has the advantage they are consistent, which I like, I like it a "LOT" I hate silicon lotteries (I lose them a lot which doesnt help), and especially when they have such a high variance. Having one chip need 1.38 to do 4.7 and another 1.25 to do 5ghz is too much variance in my opinion. A future I would like is akin to XFR, maybe PBO and all manual o/c outside of that disabled. To remove that variance, that lottery, you know what you getting for your money. Things like p-state tuning no problem, just disable the extra clocks.