AMD Ryzen 5 7600X

[bug]

You want to reuse your RAM but not your old board and CPU?

Valid - but ineffective IMO; If you reuse the old platform for yourself or friends/family you need that RAM anyway, and when I'm buying used I look for a complete platform because it's waaay better value - easier to buy a bundle and easier to sell a bundle. You could sell your CPU and board separately but it's twice the effort for you and the market for individual parts is smaller (over here at least) because each individual part incurs its own packaging and shipping costs.

I mean, if you had some DDR4 sticks, you could still use them with ADL or RL. No such luck with Zen4 though.
Sure, if you're on ADL already, you can reuse both RAM and mobo. But I've never found it worthy to upgrade from one generation to the next.

 

[Count von Schwalbe]

Good to learn something new (everyday on TPU)!!

So technically 7000-series has/had the potential of being more efficient. I thought it was primarily due to the crenels and merlons design language and the usual AMD opportunistic-volting. speaking of power consumption... are there any reviews on undervolting without harming "gaming" performance? I'd like my next upgrade to achieve a level of efficiency without having the fans ramped up, ultimately to achieve a dead-silent build without crossing the 80c mark (i understand its harmless at this range but its just one of things i can't shake off - gotto max @ ~79c or i'm willing to lose performance)

W1zzard did a small undervolt in his 7950X review - I imagine it would translate reasonably well. It actually improved performance in some cases.

 

[bug]

It's not low hanging fruit if it cuts into the capacity for the DDR5 variant. It also goes against the reusability of their silicon since now you have two I/O dies that would be special purpose essentially. One way or another, the world will move to DDR5, just as we did with DDR4 and just about everything that preceded it and I think AMD knows that. The whole advantage of the chiplet design is the same parts used for everything. You negate that advantage by doing as you suggest. There are a lot of costs and not many benefits (for AMD that is.)

To me, this is a case of AMD keeping it simple and planning for the future, which I think is the right move for the sake of DDR5 adoption.

You forget AMD already had the DDR4-capable IO dies and the fab capacity to make them. Slap on faster USB or whatever and you're done.

 

[sLowEnd]

-95c that is terrible (short life cpu?) and dont possible justify

Without any evidence that AMD has not taken proper measures to design their chips to withstand target operating temperatures, your assumption doesn't really hold water. Even as far back as the 1980s it was possible to tweak chips to tolerate various temperatures up to over 200c.

High temperature, radiation hardened electronics for application to nuclear power plants (Technical Report) | OSTI.GOV

Electronic circuits were developed and built at Sandia for many aerospace and energy systems applications. Among recent developments were high temperature electronics for geothermal well logging and radiation hardened electronics for a variety of aerospace applications. Sandia has also been...

www.osti.gov

High Temperature Semiconductors | Tekmos Inc.

Tekmos now offers products based on special high temperature processes to meet the needs of the oil & gas, aerospace, automotive and industrial markets.

www.tekmos.com

Processors today often run at temperatures way higher than they used to. My Athlon 64s and Pentium 4s loaded around mid 40c to low 50c, and IIRC max specified safe temperature was 72c for the Athlon 64. Nobody expects the same kind of behavior or limits from modern Ryzen CPUs.

 

[lexluthermiester]

Absolutely not worth to come from an overclocked Intel Core i5-12400 CPU.

I know I'm late to the party, but this made me laugh. You're the exception not the rule. Very few people are looking to upgrade from a 12th gen Intel CPU to this platform. So your statement holds about as much water as a screen door on a submarine.

 

[Valantar]

Good to learn something new (everyday on TPU)!!

So technically 7000-series has/had the potential of being more efficient. I thought it was primarily due to the crenels and merlons design language and the usual AMD opportunistic-volting. speaking of power consumption... are there any reviews on undervolting without harming "gaming" performance? I'd like my next upgrade to achieve a level of efficiency without having the fans ramped up, ultimately to achieve a dead-silent build without crossing the 80c mark (i understand its harmless at this range but its just one of things i can't shake off - gotto max @ ~79c or i'm willing to lose performance)

That's the thing though: it doesn't affect efficiency (much). Efficiency is mainly a function of the architecture, process node, voltage and clocks, and those are the same regardless of thermals - with the exception of leakage current, which increases with higher thermals. So, yes, there would be a minor power drop from less leakage current if temperature were lower - but it wouldn't be huge. Der8auer measured ~-15W with his delidded setup that reduced temperatures by 20°C, off a 170-230W CPU. So, not a huge difference. And, of course, a thinner IHS wouldn't have come close to a 20°C temperature drop. In short: a CPU is pretty much equally efficient whether it's running at 30°C or 95°C, as long as it's drawing the same power and delivering the same performance.

There have been some results posted around these threads from tests run in Eco Mode (105W and 65W at least for the 7950X, can't remember for the 7600X). Gaming performance drops were essentially nothing; productivity performance drops were more noticeable, but still much faster than the 5950X.

 

[Count von Schwalbe]

That's the thing though: it doesn't affect efficiency (much). Efficiency is mainly a function of the architecture, process node, voltage and clocks, and those are the same regardless of thermals - with the exception of leakage current, which increases with higher thermals. So, yes, there would be a minor power drop from less leakage current if temperature were lower - but it wouldn't be huge. Der8auer measured ~-15W with his delidded setup that reduced temperatures by 20°C, off a 170-230W CPU. So, not a huge difference. And, of course, a thinner IHS wouldn't have come close to a 20°C temperature drop. In short: a CPU is pretty much equally efficient whether it's running at 30°C or 95°C, as long as it's drawing the same power and delivering the same performance.

15w not much? I am typing on that much.

The V/F curve is a little high at stock. It should help stability, but this gen should undervolt well for efficiency gains.

 

[Valantar]

15w not much? I am typing on that much.

The V/F curve is a little high at stock. It should help stability, but this gen should undervolt well for efficiency gains.

It's not much out of a 170-230W CPU under full load, no. It's noticeable, sure, but again, that's with direct die cooling with liquid metal. A thinner IHS might have delivered half that temperature drop at best, and thus a much lower drop in leakage current.

 

[HTC]

Am i correctly observing these results? I'm assuming all these charts are correct based on the games employed to conduct these tests whereby some games will favour Intel over AMD and vice versa.

Please shed some light!

Dunno if anyone answered (haven't read all the replies) but i think i can answer: it's the coolers used.

Correct me if i'm wrong: isn't this the 1st time EVER a CPU's stock performance is so directly tied to the cooler used?

Since the CPU will boost as high as it can possibly get WHILE the temp doesn't surpass 95º, a review that used a stronger cooler will see better results in performance than a review that used a weaker one. This means that YOU CAN'T COMPARE different reviews UNLESS the same cooler model was used for both.

 

[Why_Me]

Dunno if anyone answered (haven't read all the replies) but i think i can answer: it's the coolers used.

Correct me if i'm wrong: isn't this the 1st time EVER a CPU's stock performance is so directly tied to the cooler used?

Since the CPU will boost as high as it can possibly get WHILE the temp doesn't surpass 95º, a review that used a stronger cooler will see better results in performance than a review that used a weaker one. This means that YOU CAN'T COMPARE different reviews UNLESS the same cooler model was used for both.

AMD Ryzen 5 7600X Review - Affordable Zen 4 for Gaming

Ryzen 5 7600X is the company's most affordable offering for the Zen 4 family. Our review of the 7600X confirms that this new $300 CPU offers huge performance gains over Zen 3, and can even beat the Ryzen 5800X3D in gaming. In applications, the 7600X is faster than the Intel Core i9-11900K.

www.techpowerup.com

Air Cooling: Noctua NH-U14S
Water Cooling: Arctic Liquid Freezer II 420mm

 

[HTC]

AMD Ryzen 5 7600X Review - Affordable Zen 4 for Gaming

Ryzen 5 7600X is the company's most affordable offering for the Zen 4 family. Our review of the 7600X confirms that this new $300 CPU offers huge performance gains over Zen 3, and can even beat the Ryzen 5800X3D in gaming. In applications, the 7600X is faster than the Intel Core i9-11900K.

www.techpowerup.com

Air Cooling: Noctua NH-U14S
Water Cooling: Arctic Liquid Freezer II 420mm

I wonder what the STOCK performance would be if someone tried to use ... a chiller ... or something even stronger.

I also wonder how much performance "would be lost" by using ... say ... a Cooler Master Hyper 212 120mm, for example.

It would be interesting to see how different the performance would be.

 

[W1zzard]

I wonder what the STOCK performance would be if someone tried to use ... a chiller ... or something even stronger.

Almost none, the 7950X review has some testing with AIO+undervolt, and the deltas are tiny

I also wonder how much performance "would be lost" by using ... say ... a Cooler Master Hyper 212 120mm, for example.

Planning for such an article right now. Run our Noctua at various fan speeds and record temp, clocks, perf in a few apps and perf in a few games

 

[HTC]

and record temp, clocks, perf in a few apps and perf in a few games

Efficiency too, please.

@ least lowest fan speed VS highest fan speed: if that doesn't produce enough efficiency variance, then there's no point in testing the "middle fan speed(s)".

 

[W1zzard]

least lowest fan speed

I was thinking 0% fan speed as minimum ^^

 

[wheresmycar]

Almost none, the 7950X review has some testing with AIO+undervolt, and the deltas are tiny


Planning for such an article right now. Run our Noctua at various fan speeds and record temp, clocks, perf in a few apps and perf in a few games

I was thinking 0% fan speed as minimum ^^

If you're running these tests with a 7600X, in the process it would be amazing if you could undervolt the 7600X and see what you get. You are the W1zzard hence anything is possible

That's the thing though: it doesn't affect efficiency (much). Efficiency is mainly a function of the architecture, process node, voltage and clocks, and those are the same regardless of thermals - with the exception of leakage current, which increases with higher thermals. So, yes, there would be a minor power drop from less leakage current if temperature were lower - but it wouldn't be huge. Der8auer measured ~-15W with his delidded setup that reduced temperatures by 20°C, off a 170-230W CPU. So, not a huge difference. And, of course, a thinner IHS wouldn't have come close to a 20°C temperature drop. In short: a CPU is pretty much equally efficient whether it's running at 30°C or 95°C, as long as it's drawing the same power and delivering the same performance.

There have been some results posted around these threads from tests run in Eco Mode (105W and 65W at least for the 7950X, can't remember for the 7600X). Gaming performance drops were essentially nothing; productivity performance drops were more noticeable, but still much faster than the 5950X.

Yeah i have bad habit of referring to thermals as "efficiency"... can't count on me fingers how many times i've been corrected on that lol To compensate follow up unavoidable errors, maybe i should reinvent temps as "thermal efficiency" - if the tech community hasn't already trademarked these terms to define something else.

Valantar, i recall in some thread you mentioned B-series boards will be around $150/$180+ (maybe it was someone else?).... was that speculation or officially reported "starting from" prices. The X-series, with a limited selection of boards currently available, start from £350 in the UK. I refuse to fork out anything above £250, preferably £150-£200 for a basic non-compromising build for gaming and basic office/personal use. Just hope these not-so feature rich $150+ boards will be up-to-the-task for ~2025+ forward Gen support.

 

[Valantar]

Valantar, i recall in some thread you mentioned B-series boards will be around $150/$180+ (maybe it was someone else?).... was that speculation or officially reported "starting from" prices. The X-series, with a limited selection of boards currently available, start from £350 in the UK. I refuse to fork out anything above £250, preferably £150-£200 for a basic non-compromising build for gaming and basic office/personal use. Just hope these not-so feature rich $150+ boards will be up-to-the-task for ~2025+ forward Gen support.

I haven't said anything about specific prices (I don't know anything about that, unfortunately ), but I might have speculated about base prices maybe? I know I wrote something about something in that direction in the past couple of days, but I can't remember if I pulled a number out of my rear end or if I just said "cheap" You might be thinking about someone else too I guess.

I really, really hope we'll see decent B650 boards in that range as well, and I completely agree with you on 250+ boards being just silly for pretty much any ordinary use. They have tons of features, so it's not like you aren't getting anything for your money - but it's not things that people actually need. Personally I wouldn't worry about forward compatibility though. If it has VRMs to run a 7950X full tilt (which it ought to have - but that's also a big part of why prices are rising!) then it should handle any future AM5 CPU, given that the socket is specced at 230W, and that is unlikely to change. Of course you're still dependent on BIOS updates to support new hardware. I just hope AMD is less flaky about that this time around.

 

[wheresmycar]

I haven't said anything about specific prices (I don't know anything about that, unfortunately ), but I might have speculated about base prices maybe? I know I wrote something about something in that direction in the past couple of days, but I can't remember if I pulled a number out of my rear end or if I just said "cheap" You might be thinking about someone else too I guess.

I really, really hope we'll see decent B650 boards in that range as well, and I completely agree with you on 250+ boards being just silly for pretty much any ordinary use. They have tons of features, so it's not like you aren't getting anything for your money - but it's not things that people actually need. Personally I wouldn't worry about forward compatibility though. If it has VRMs to run a 7950X full tilt (which it ought to have - but that's also a big part of why prices are rising!) then it should handle any future AM5 CPU, given that the socket is specced at 230W, and that is unlikely to change. Of course you're still dependent on BIOS updates to support new hardware. I just hope AMD is less flaky about that this time around.

Ooops might have been someone else.

I hope it holds weight though with barebone options starting from $150 and the finer clothed offerings somewhere in the $200 region. But, my goodness £350 for a starting point for X series adds doubt. I totally agree, most of these hi-end features are not of any use for most people (or a major "some") and I can't see that changing in the 3-year+ support cycle.

Is there a particular reason why B-series is always launched after the premium stuff? Same with non-X variant processors? Is it down to profitability with higher priced parts or early release competitiveness with best available hardware? I see the same thing with Intel... the pricier stuff always first.

 

[Count von Schwalbe]

I may have stated that B650 will run around $150 to start with after someone talked about platform cost. I then looked at AMD's official marketing slides, and IIRC they were saying $120, and a launch in October.

I see the same thing with Intel... the pricier stuff always first.

Who doesn't do that?

 

[Valantar]

Ooops might have been someone else.

I hope it holds weight though with barebone options starting from $150 and the finer clothed offerings somewhere in the $200 region. But, my goodness £350 for a starting point for X series adds doubt. I totally agree, most of these hi-end features are not of any use for most people (or a major "some") and I can't see that changing in the 3-year+ support cycle.

Is there a particular reason why B-series is always launched after the premium stuff? Same with non-X variant processors? Is it down to profitability with higher priced parts or early release competitiveness with best available hardware? I see the same thing with Intel... the pricier stuff always first.

Yep, pretty much. They want to recoup as much of their R&D and early production costs as quickly as possible, so they only sell premium SKUs in the beginning. If you're impatient, you pay for the privilege; if you're patient you inevitably get better value.

 

[wheresmycar]

I may have stated that B650 will run around $150 to start with after someone talked about platform cost. I then looked at AMD's official marketing slides, and IIRC they were saying $120, and a launch in October.

ahh we found the culprit lol...

$120 for B-series? i doubt it. Maybe A-series? Around $150 for B-series would be triumph for the consumer

Who doesn't do that?

I'm just curious why that was... if its profitability alone, i guess it makes sense from the seller/manufacturers perspective. As an inquisitive consumer, it would be nice if they could simply drop all the variants in one go... first we get informed about Next Gen (we wait), then we get informed of a date of announcement (we wait), then we wait for the announcement to confirm the release date (we wait) and finally when the stuff is released which is of the premium variety we then look forward to the more reasonable offerings (we wait some more). So i'm just curious why the intrusive teasing... the upgrade itch, forget the skin, is cutting thru me bones!

Yep, pretty much. They want to recoup as much of their R&D and early production costs as quickly as possible, so they only sell premium SKUs in the beginning. If you're impatient, you pay for the privilege; if you're patient you inevitably get better value.

I thought so. Can't blame them seeing how people are willing to pay their skin off for small performance gains (esp in the GPU sector)

 

[Count von Schwalbe]

$120 for B-series? i doubt it. Maybe A-series? Around $150 for B-series would be triumph for the consumer

B550 with full PCIe 4.0 support can be had for under $100. No A-series coming. AMD is only making one chipset this gen - Promontory 21. B650 is a single one of these, and X670 is two. The E versions are just certification for PCIe 5.0 on the GPU slot, so nothing actually to do with the chipset.

The only real price increase from B550 to B650 should be the M.2 and DMI being PCIe 5, potential socket cost increase, and DDR5 traces.

 

[Valantar]

B550 with full PCIe 4.0 support can be had for under $100. No A-series coming. AMD is only making one chipset this gen - Promontory 21. B650 is a single one of these, and X670 is two. The E versions are just certification for PCIe 5.0 on the GPU slot, so nothing actually to do with the chipset.

The only real price increase from B550 to B650 should be the M.2 and DMI being PCIe 5, potential socket cost increase, and DDR5 traces.

Yes, but B550 has also been on the market for a year and a half. Prices drop. Still, B650 shouldn't be that much more expensive than B550, outside of VRM costs. Hopefully we'll see some sensible boards.

 

[HTC]

I was thinking 0% fan speed as minimum ^^

I was thinking around 300 RMP or so ...

 

[Count von Schwalbe]

Yes, but B550 has also been on the market for a year and a half. Prices drop. Still, B650 shouldn't be that much more expensive than B550, outside of VRM costs. Hopefully we'll see some sensible boards.

Just checked price history, 2 years ago the DS3H was $95. Some people just make cheap boards.

 

[catulitechup]

ahh we found the culprit lol...

$120 for B-series? i doubt it. Maybe A-series? Around $150 for B-series would be triumph for the consumer

Yes, but B550 has also been on the market for a year and a half. Prices drop. Still, B650 shouldn't be that much more expensive than B550, outside of VRM costs. Hopefully we'll see some sensible boards.

90 to 95 degrees on 120us mainboard (if appear, personally wait for 150us and beyond for entry b650)

resuming this a big flop from amd but dont as bad like bulldozer, wait no this is worst than bulldozer because bulldozer support ddr3 but zen 4 have higher temperatures maybe caused to try show zen 4 are better than raptorlake up frecuencies* (however seems them possible fail in same cpu area like i5 13600k vs ryzen 5 7600x)

*if them use some lower frecuencies must be have better temps for example turbo at 5.0ghz or 5.1ghz must be have better temps
and now thanks to amd now delidded must be considered on zen 4 with higher frecuencies, in this video der8auer shows zen 4 delidd and them stay preparing delidd kit

and use ddr5 (expensive now and have some cheap sticks but with suck frecuencies and latencies like 4800/5200mhz kits) only for now and for see differences need high speed ddr5 memory aka 6000mhz and beyond