AMD Ryzen 7 7700

[ir_cow]

Will this problem get fixed ?
Is this bound with the Motehrboard or something ? The others brands do it better ?
I'm asking, because it's really great to have a PC starting in les than 10s these days, faster than a PS4 ...
When i'm reading +30s at launch, i fell like a big step back ...

Its not a bug, its the way it has been designed. memory training of the sub-timings gives a perf boost.

 

[Vario]

I am also curious about the DDR5 slow boots for Zen 4 at launch versus how it stands presently.

 

[TumbleGeorge]

I am also curious about the DDR5 slow boots for Zen 4 at launch versus how it stands presently.

As already described in a previous comment. +20-60 seconds. Which is better than before few months around release date of Ryzen 7000X series.

 

[Arco]

As already described in a previous comment. +20-60 seconds. Which is better than before few months around release date of Ryzen 7000X series.

Weird, I've had great boot times. (10 seconds)

 

[ir_cow]

Weird, I've had great boot times. (10 seconds)

Which MB and are you enabling XMP/EXPO?

 

[Arco]

Which MB and are you enabling XMP/EXPO?

Asrock Steel Legend X670E, Yes.

 

[Kaleid]

With the latest bios I now load as fast into windows as with 5800x3d. Using 7600x

 

[cbb]

ok, dumb ? time; it sure looks like the 4k gaming tests are largely gpu bottlenecked...making the comparison kinda useless? am I reading it wrong? (prolly!) if not, maybe use biggest avail gpu for testing cpus, to show how they'll fare in future as gpus improve?

That said, great reviews; this is pretty much my go to site these days (as others have sold out/fallen by the wayside). Further, this is looking like my likely new cpu (have old 4790k/2070); this may be good enough until there's a gpu at a reasonable price to replace my 2070 (i.e. not anytime soon, probably...sigh); "good enough" to match my current gpu and even a 2-3gen newer replacement when that happens. If I'm reading all this correctly (ponders orb carefully!)

 

[AnarchoPrimitiv]

Well, just goes to show you how dumb the maxed out Wattage on the X CPUs was.

50W savings in MT, 20C lower temps and only 5% performance loss in worst case scenarios.

Yes, but the vast majority of consumers don't think about such things....they might look at a few pcmag reviews, or maybe not even that, and then they just buy what some idiot they know tells them to buy

I see my i7 12700K is still one of the best gaming CPU's out there, I'm all good...

Declaring that aloud makes me doubt your confidence. Haha

 

[Hofnaerrchen]

Being teased by a quite short review from a different tech outlet I am really looking forward to the R9 7900 review. Based on what I've seen so far it's efficiency is even better than that of the R7 7700.

ok, dumb ? time; it sure looks like the 4k gaming tests are largely gpu bottlenecked...making the comparison kinda useless? am I reading it wrong? (prolly!) if not, maybe use biggest avail gpu for testing cpus, to show how they'll fare in future as gpus improve?

That said, great reviews; this is pretty much my go to site these days (as others have sold out/fallen by the wayside). Further, this is looking like my likely new cpu (have old 4790k/2070); this may be good enough until there's a gpu at a reasonable price to replace my 2070 (i.e. not anytime soon, probably...sigh); "good enough" to match my current gpu and even a 2-3gen newer replacement when that happens. If I'm reading all this correctly (ponders orb carefully!)

For gaming in high resolutions you can save quite some money by buying a cheaper CPU (and memory) and putting the saved money in an even better GPU. That's the case since the release of the first gaming GPU - with some small caveats.

 

[AusWolf]

Thanks for the review!

So you had to dial up the voltage to 1.3 V for overclocking. That's very interesting considering that the 7700X runs at 1.4 V by default, and only boosts 200 MHz higher in single-core work.

So, the 7700X vs. 7700 was an exact rehash of the 5800X vs. 5700X, and there's zero reason why Zen 4 can't have equally manageable thermals as Zen 3 without dipping below 5GHz. Way too much PPT + (maybe) high Vcore to compensate for lower quality early N5 silicon = inferno on 7700X

Look at the all-core sustained clocks - 7700 is barely behind by 100MHz, a difference that can probably even be overcome purely with a larger Curve Optimizer offset (which doesn't affect thermals at the same PPT).

AMD tried to justify it with their horseshit "95C is the new normal, it boosts opportunistically", when in reality Precision Boost still functions the same way it did on Zen 3 and the excuse was just to cover for their pricing decisions/early production SP.


Improved V-F bodes well for the X3D parts though. We're what, 3 months past launch?

I have a feeling that I could achieve similar results with my X just by setting a lower PPT or temperature target, despite the difference in max. voltage. But as long as it behaves similarly in games, I'm not bothered.

 

[tabascosauz]

So you had to dial up the voltage to 1.3 V for overclocking. That's very interesting considering that the 7700X runs at 1.4 V by default, and only boosts 200 MHz higher in single-core work.

1.4V during all-core workloads?? That's pretty crazy, I don't think anything since Pinnacle has come close to 1.4V stock at max load.

 

[ir_cow]

Asrock Steel Legend X670E, Yes.

Strange.... Just reviewed the B650E it was quicker than others, but not 10 seconds as your claiming. Is this is from a off state or sleep mode?

I'm not quite sure how these could be so different, but memory configurations as endless. My gut says either you have it as ASUS version of context restore enabled, which could be by the fault now (for all I know ) with the newest bios or you're just coming out of sleep mode (waking up the computer) and computers actually not off off.

It's not that I don't believe you totally, I just think it's odd and not the norm.

 

[W1zzard]

maybe use biggest avail gpu for testing cpus,

3080 is still "fast" I'd say, but yes, next CPU testing rig will use 4090

So you had to dial up the voltage to 1.3 V for overclocking

Technically not "up" but "set to fixed". By default the voltage varies big time depending on many factors like number of cores active

 

[AusWolf]

1.4V during all-core workloads?? That's pretty crazy, I don't think anything since Pinnacle has come close to 1.4V stock at max load.

Oh no! 1.4 is the maximum during maximum boost in single-threaded loads (with 5550 MHz on those threads).

Technically not "up" but "set to fixed". By default the voltage varies big time depending on many factors like number of cores active

I see. Do you know what its default maximum value is?

 

[W1zzard]

I see. Do you know what its maximum value is?

I didn't check, and it'll vary anyway due to silicon lottersy

 

[AusWolf]

I didn't check, and it'll vary anyway due to silicon lottersy

A good point.

 

[xorbe]

Is the slow boot everytime, or just the first time?

 

[mechtech]

Why the actual f**k did AMD not lead the AM5 launch with these CPUs instead of the X parts?

Oh right, greed, that's it. And look where that got them.

Because nvidia and Intel launch with top end parts so they follow suit???

 

[AusWolf]

Is the slow boot everytime, or just the first time?

It's around 30 s - 1 min (depending on your memory configuration) every time. It doesn't bother me personally.

 

[DemonicRyzen666]

There is a typo in the conclusion on the lists

Intel / 7700X

you mean 13700K ?

 

[Minus Infinity]

so sad to see how AMD had to play Intel's marketing games and push the X cpu's to the max for hardly any gain but a lot more power and heat. The performance loss is minimal even at 65W. Honestly I'd just run something like 95W (if possible) for productivity and leave it at 65w for gaming. Just check out Anandtech's power scaling of the 7950X vs 13900X. If you already have an X cpu just power limit it.

I now wish the v-cache models were non-x as well.

 

[AusWolf]

so sad to see how AMD had to play Intel's marketing games and push the X cpu's to the max for hardly any gain but a lot more power and heat. The performance loss is minimal even at 65W. Honestly I'd just run something like 95W (if possible) for productivity and leave it at 65w for gaming. Just check out Anandtech's power scaling of the 7950X vs 13900X. If you already have an X cpu just power limit it.

I now wish the v-cache models were non-x as well.

I agree, although you only see crazy power and temperature in all-core workloads. If you only game, it doesn't matter.

 

[RJARRRPCGP]

Asrock Steel Legend X670E, Yes.

Is it just me, or are those ASRocks looking very good lately, including their B550s and X570 Steel Legends?

 

[SchumannFrequency]

I've been complaining about extremely long boot times in my original Zen 4 reviews, and AMD assured us that these are fixed. To my surprise nothing was fixed and the new 65 W CPU models took just as long to boot—30 seconds or more—every single time. Turns out that on ASUS motherboards you need to enable the "Memory Context Restore" BIOS option, which saves some memory training info after the first attempt and reuses that on subsequent reboots. Kinda dumb that the option is turned off by default, even on the latest 0805 BIOS from last month. With "Memory Context Restore" enabled, boot times are still longer than on other platforms, but only by a few seconds and are now in a range that I would call "acceptable."

Will this problem get fixed ?
Is this bound with the Motehrboard or something ? The others brands do it better ?
I'm asking, because it's really great to have a PC starting in les than 10s these days, faster than a PS4 ...
When i'm reading +30s at launch, i fell like a big step back ...

I can safely say that this has nothing to do with the CPU. And it is mainly caused by the motherboard or other hardware or in addition perhaps also a wrong setting (in windows or the bios).

I have this hardware: Intel i3-3240 @ 3.392GHz + 4GB RAM @1600MHZ single channel + NVIDIA GTX 650 1GB + EVO 850 500GB

The BIOS takes about 13 seconds, but once GRUB is loaded it takes 2.9 seconds to login according to systemd (on this weak hardware). This means that the operating system itself only needs 2.9 seconds to go to the login. And that's systemd, which isn't the fastest init system in Linux. S6 init is the fastest, about twice as fast as systemd. So if Clear Linux were to switch to S6 for the init my entire operating system could boot in less than 2 seconds. (on my weak hardware)

You see what my hardware is so that's pretty spectacular. Void Linux also has extremely fast boot times. DragonFly BSD also boots up quickly.
If you're using Windows, there's an option that uses hibernation techniques to make your computer boot faster. This does have a negative impact on driver stability.