AMD Ryzen 7 3800XT

[TheoneandonlyMrK]

"What was AMD thinking pricing this thing at $399?"

It's the oldest marketing trick in the world. Inflate prices, then soon release a new better product at the same price and everyone is happy so you can keep higher profit margin. Plus you don't have to discount previous generation of products too much. That's exactly what NGreedia did with release of Super GPUs in Turing line. Expect these prices to translate into 4000 line.

Something has to be near the top, you can get cheaper.

 

[Shatun_Bear]

Is there actually people who would buy a 3800XT when they can have 10700K for the same price? Hard to imagine

The 10700K MRRP is $375 but I can't see it at that price anywhere, Newegg has them for $410, Amazon there's no stock.

Also, aren't Intel motherboards more expensive too? Last small point is the 10700K draws a hell of a lot more power under general load.

 

[Fluffmeister]

Wow, and to think... these are the "good guys" in the duopoly.

 

[Gmr_Chick]

The 10700K MRRP is $375 but I can't see it at that price anywhere, Newegg has them for $410, Amazon there's no stock.

Also, aren't Intel motherboards more expensive too? Last small point is the 10700K draws a hell of a lot more power under general load.

I (somehow) managed to snag my 10700K off Amazon for $374, no shit. It's been out of stock ever since there, or sold by 3rd parties for the price of a kidney, or the life of your first-born child, lol.

Also, since AMD dropped B550, and you can't find X570 boards in stock, this whole "Intel's boards are more expensive" delusion has to die already. Yes, some higher-end Z490 boards are crazy expensive, but so are some X570 boards, and a lot of B550 boards are grossly over-priced. On the other hand, Intel has boards up and down the chipset stack priced accordingly.

 

[R0H1T]

Well Intel doesn't need to charge a premium for the right privilege to OC a chip, the segmentation bit is true wrt AMD though you do know that x570 is the best chipset in AMD's lineup for a number of reasons not just related to OCing.

On the other hand, Intel has boards up and down the chipset stack priced accordingly.

Intel's still operating as if they're the only game in town, they need to scrap their BS market segmentation. Their chip segmentation is literally ridonculous though chipset segmentation is relatively less of a hassle when you look at their CPU stack

 

[r9]

I always cheer for the underdog, but since when AMD are the A side ?! They are beaten in productivity, gaming and price. Let me summarize it's crap.

 

[cucker tarlson]

Is there actually people who would buy a 3800XT when they can have 10700K for the same price? Hard to imagine

man,where you been the last couple of years ?

Also, aren't Intel motherboards more expensive too? Last small point is the 10700K draws a hell of a lot more power under general load.

no,it's the other way round

idle,light load,single thread,gaming - ryzen 3000 draws more.
intel goes beserk on stress tests

 

[Palladium]

I wonder just how brain damaged does one have to be to pick this over the 3900X?

 

[cucker tarlson]

The 10700K MRRP is $375 but I can't see it at that price anywhere, Newegg has them for $410, Amazon there's no stock.

Also, aren't Intel motherboards more expensive too? Last small point is the 10700K draws a hell of a lot more power under general load.

10700kf is the direct competitor

 

[Pumper]

This trash is 10€ more expensive than 3900X on mindfactory. What a joke.

 

[argon]

Is there actually people who would buy a 3800XT when they can have 10700K for the same price? Hard to imagine

I see what you did there, and you did it right.

btw,
tpu , what about cod warzone benchmark is gonna be very interesting!!!!! we need that piece of crap game on stats, also RDR2 would be interesting.

 

[W1zzard]

cod warzone

Always online junk, and no single player, so no way to get proper reproducable results .. no thanks

RDR2

I already hate having RDR2 in my GPU bench, no plans to make my CPU benching life miserable, too

 

[Assimilator]

intel goes beserk on stress tests

Or overclocking.

I already hate having RDR2 in my GPU bench, no plans to make my CPU benching life miserable, too

Rockstar quality game engines and console ports, amirite?

If RDR2 is giving you so much grief remove it, and state the reasons for doing so, and when people cry about it tell them to cry to Rockstar instead.

 

[cucker tarlson]

Or overclocking.

exactly
it's why it's better to stick to non-K's this time round
you're getting 92-95% the performance at surprisingly good power
once you push for 5GHz allcore kiss cool and quiet goodbye

 

[HenrySomeone]

Or overclocking.

Actually, they only go truly berserk when both OCed and under stress tests; in all other scenarios, it's manageable.
In gaming, fully OCed 10700k will only push the system consumption up by a measly 5%

 

[Assimilator]

exactly
it's why it's better to stick to non-K's this time round
you're getting 92-95% the performance at surprisingly good power
once you push for 5GHz allcore kiss cool and quiet goodbye

It's such an incredible irony that the only company that makes CPUs that can actually overclock worth a damn, is also the company that can't get their damn fabrication process right so that overclocking those CPUs to high levels is actually something you can consider without cooling sufficient to freeze the Sun.

I wonder if AMD is going to lift some limits with Zen 3 to allow users to push whatever wattage they desire through their chips. At least then the unnecessary extra 4-pin and 8-pin power connectors on some of the high-end boards would finally actually have a use.

Actually, they only go truly berserk when both OCed and under stress tests; in all other scenarios, it's manageable.
In gaming, fully OCed 10700k will only push the system consumption up by a measly 5%

I wish people would stop cherry-picking performance numbers, especially on these latest Intel chips where the power draw and heat output begin to approach exponential levels as the clock is ramped up.

The full context is that 5% more power equates to a "measly" 27% more heat over max turbo...

 

[HenrySomeone]

Not in gaming though:
"We used a Noctua NH-U12 to measure CPU temperature while running Blender."
Although generally I agree that 10700 and 10900 series in particular give superb performance out of the box already.

 

[Assimilator]

Not in gaming though:
"We used a Noctua NH-U12 to measure CPU temperature while running Blender."
Although generally I agree that 10700 and 10900 series in particular give superb performance out of the box already.

Hmmm, I think the bigger issue here is that the temperature test is completely independent of his power consumption tests, hence there's no way to correlate the two metrics. Which really makes the temperature test useful only as a measure of the absolute maximum temperatures you're likely to see.

Suggestion to @W1zzard:

• additional "Power Consumption" test graph for Blender
• additional temperature graphs for each of the current "Power Consumption" test cases (Idle/Single-threaded/Multi-threaded/Stress test/Gaming) - measuring the max temp recorded over the run (not necessarily the temp at the time of max power draw)

This would then allow users to compare and correlate power against temperature over these different workloads. So for example, if you're interested in building a PC only for gaming you'd be able to see the max power and temps that you should expect, which would help you to plan for buying PSU and cooler respectively.

I know this would be more work for you, but I do believe it would add a ton more value to these reviews.

 

[W1zzard]

Blender power should be nearly identical to Cinebench power.

Good suggestion though for next test system. If you had to pick one for both, which would you choose and why?

 

[Assimilator]

Blender power should be nearly identical to Cinebench power.

Good suggestion though for next test system. If you had to pick one for both, which would you choose and why?

One of the workloads, you mean? I personally would pick whatever is the worst-case (highest power/temp), but of course that would require you to record temp on every one of the 5 power scenarios you currently provide. If I absolutely had to choose one of the current 5, I'd choose gaming because that is likely the use-case that the majority of review readers would be interested in.

If you want to keep it simple, you could simply replace the current Blender temperature test graph with a Cinebench temp test one.

 

[W1zzard]

One of the workloads, you mean?

I don't want to add yet another test, especially since they are so similar.

So either:
Temps & Power = Cinebench
or
Temps & Power = Blender

I slightly lean towards Blender because that's more real-life

 

[Turmania]

So is this worth it or wait for ryzen 3?

 

[Renald]

Is there actually people who would buy a 3800XT when they can have 10700K for the same price? Hard to imagine

I wouldn't, because of the energy consumption. I live in a pretty hot place, and in summer, I have to chose between having +5FPS 105W or a 65W TDP (mine is a 3700X) CPU. Choice is made easily.
Of course these 3800XT and such are not attractive at all, but that doesn't make 10700k viable either. And 10700k consumption is worse.

I think that's his point. A bad product doesn't make his slightly better twin a good thing. It's just a lesser failure.

 

[Gmr_Chick]

Not in gaming though:
"We used a Noctua NH-U12 to measure CPU temperature while running Blender."
Although generally I agree that 10700 and 10900 series in particular give superb performance out of the box already.

Totally agree with this! Despite being a K SKU, I don't even really feel the need to OC my 10700K, as I find its performance at stock to be sufficient for me!

 

[cucker tarlson]

Totally agree with this! Despite being a K SKU, I don't even really feel the need to OC my 10700K, as I find its performance at stock to be sufficient for me!

4.7 allcore is pretty high already.
imo do stock voltage +0.05v and see where it goes.though your aio should handle 5G alright.just not worth it for the power output imo.

I wouldn't, because of the energy consumption. I live in a pretty hot place, and in summer, I have to chose between having +5FPS 105W or a 65W TDP (mine is a 3700X) CPU. Choice is made easily.
Of course these 3800XT and such are not attractive at all, but that doesn't make 10700k viable either. And 10700k consumption is worse.

I think that's his point. A bad product doesn't make his slightly better twin a good thing. It's just a lesser failure.

really ? do you stress tests all the time ?
cause idle,single threaded,light load,gaming intel takes less power