Question Will CPU supplies return? Ryzen 2700 (or replacement?)

[asking]

Greetings,

I've been patiently waiting out the pandemic and watching ridiculous price gouging on AMD Ryzen 2700 CPUs. Not all of the AMD line seems affected, so it's hard to know if one should keep waiting or if the 2700 is just being phased out.

Any insights on that, people?

I'd love to buy a newer Ryzen, but the Anandtech reviews of everything newer show that actual power usage on Ryzen 2 CPUs is way over advertised TDP, but the 2700 really does hover at or even below 65W. Is there any other CPU that can compete at this level of (honest) power consumption?

 

[Rigg]

Micro Center rules. I bought 4 2700x CPU's for $130 each last fall. I picked up great bundle deals on new x370 strix boards they were blowing out with them too. I've gotten a handful of amazing open box deals there over the years. I grabbed an RX 470 8 gig for $13 and a 1070 ti for $170 last year. I've picked up multiple b450 boards for less than $25 with bundles.

The store is right by my office so they tend to get a lot of my disposable income.

 

[scannall]

Micro Center rules. I bought 4 2700x CPU's for $130 each last fall. I picked up great bundle deals on new x370 strix boards they were blowing out with them too. I've gotten a handful of amazing open box deals there over the years. I grabbed an RX 470 8 gig for $13 and a 1070 ti for $170 last year. I've picked up multiple b450 boards for less than $25 with bundles.

The store is right by my office so they tend to get a lot of my disposable income.

In some ways I'm glad the closest one to me is 1400 miles...

 

[ElFenix]

RX 470 8 gig for $13

 

[Yeroon]

I put my watt meter back on my 3600 system, and the only way it stays close to the 65w rated tdp when doing avx2 style workloads is CPB off, base frequency (3.6ghz).
Idle 40w wall, CPB off, ram at 3200, y-cruncher pulled 119w peak. PSU is 450w 80+ gold.

All stock bios (well disabled the rgb eye cancer) CPB on, peak hit 152w (edit: while running/boost throttling at 95*C)

I'm guessing a 3700x would require similar settings (CPB disabled, base clock freq) to adhere to its 65w rated TDP under avx2 type loads.

Edit: while light loads saw 4.2ish ghz with cpb on, loaded up with y-cruncher it mostly sat in the 3.75-3.85ghz range, 1.25-1.3v. Temps were also up 25-30* when using a NH-L9a in open air vs base clocks.

 

[rbk123]

I'm guessing a 3700x would require similar settings (CPB disabled, base clock freq) to adhere to its 65w rated TDP under avx2 type loads.

I have no doubt this is true, however if you did the same workload (i.e. Handbrake), albeit without avx2, on the 2700 it will also go well beyond the 65W TDP rating. That's the point Hitman was trying to make.

But the OP knows that. He wasn't really looking for assistance...

 

[Yeroon]

I have no doubt this is true, however if you did the same workload (i.e. Handbrake), albeit without avx2, on the 2700 it will also go well beyond the 65W TDP rating. That's the point Hitman was trying to make.

But the OP knows that. He wasn't really looking for assistance...

False, but why bother looking it up when you can just state opinion.

Lowest power draw from the wall was the 2700 in blender, at 126w:

AMD's Ryzen 7 2700 CPU reviewed

techreport.com

I understand the setup is different, however mine is pretty well optimized for high eff at low power draws. So for a 2700 to end up with comparable wall draw vs my 3600 at its base, tells me it mostly adheres to its rated 65w tdp.
65w 3000 series at default violates its rated tdp much more so than a 65w 2700. And there are use cases where the power draw is the limit.

I also understand the zen2 cores would have a higher performance for the same wattage if the bios was set for it to maintain its rated tdp. A 3700x with CPB disabled is still going to outperform a 2700, and likely stay close to 65w, but thats not covered in most reviews.

 

[blckgrffn]

False, but why bother looking it up when you can just state opinion.

Lowest power draw from the wall was the 2700 in blender, at 126w:

AMD's Ryzen 7 2700 CPU reviewed

techreport.com

I understand the setup is different, however mine is pretty well optimized for high eff at low power draws. So for a 2700 to end up with comparable wall draw vs my 3600 at its base, tells me it mostly adheres to its rated 65w tdp.
65w 3000 series at default violates its rated tdp much more so than a 65w 2700. And there are use cases where the power draw is the limit.

I also understand the zen2 cores would have a higher performance for the same wattage if the bios was set for it to maintain its rated tdp. A 3700x with CPB disabled is still going to outperform a 2700, and likely stay close to 65w, but thats not covered in most reviews.

What is default though? Default in that particular BIOS an given board? With the memory profiles on or off? Etc.

I think what has frustrated many others on this forum is that many motherboard manufacturers are way more biased towards getting review points than they are strictly toeing the line of the "default" CPU configuration. At what point do we have to have reviews that go through and state every single option that was specified in the bios when conducting the review? And then they change them to see what happens? And then have them do that with a large enough sample so it is really statistically significant?

That's to even attempt to account for differences within the same CPU just because some are better silicon than others.

It would be science if you, with your setup, were to have a 2700, set your bios with the same settings across the board with both CPUs (PBO on/off, etc.) and take measurements with your meter running the same tests.

Beyond that, I mean, it's horseshoes & hand grenades out there.

 

[Yeroon]

What is default though? Default in that particular BIOS an given board? With the memory profiles on or off? Etc.

I think what has frustrated many others on this forum is that many motherboard manufacturers are way more biased towards getting review points than they are strictly toeing the line of the "default" CPU configuration. At what point do we have to have reviews that go through and state every single option that was specified in the bios when conducting the review? And then they change them to see what happens? And then have them do that with a large enough sample so it is really statistically significant?

That's to even attempt to account for differences within the same CPU just because some are better silicon than others.

It would be science if you, with your setup, were to have a 2700, set your bios with the same settings across the board with both CPUs (PBO on/off, etc.) and take measurements with your meter running the same tests.

Beyond that, I mean, it's horseshoes & hand grenades out there.

It'd be nice if I had a 2700 to do a back to back comparison. However, I only have access to my 3600 and possibly a 3300x machine with similar specs.

For the 3 b450 boards I have used this year, (asus b450-i Strix, B450M-tuf, asrock b450m pro4) CPB was enabled in cpu sub menus. I believe this is how 3000 series is marketed, and therefore it makes sense that it is enabled by default. Not sure how at that point AMD can also claim the TDP they do, but its not like that isn't already a pretty useless number with how intel has been treating it.

Eco mode not being a bios option is a bit of a let down as well. I was unable to find any settings to specify tdp anywhere in the bios, and as a linux user, there is no software afterwards for me to lean on to limit power.

 

[rbk123]

False, but why bother looking it up when you can just state opinion.

Lowest power draw from the wall was the 2700 in blender, at 126w:

False. But why bother comparing apples to apples when you have a singular data point that doesn't describe the setup details we've been discussing in this thread. Try again.

Note, my original post wasn't directed at you, but rather agreeing with your assumptions on the 3700X power draw. And that those assumptions should also be applied to the 2700 as has been mentioned multiple times in this thread.

 

[Yeroon]

Note, my original post wasn't directed at you, but rather agreeing with your assumptions on the 3700X power draw. And that those assumptions should also be applied to the 2700 as has been mentioned multiple times in this thread.

But the data available shows the 2700 does not behave like a 2700x or 3700x when it comes to power usage with default settings.

2700x did go well over its 95w, that was pretty well known to use its 142w cap unless it hit other limits. The review I linked shows that in the wall power consumption as well.

Heres another one:

AMD Ryzen 7 2700 8-Core 65W Processor Review - Page 8 of 10 - Legit Reviews

- Page 8 of 10

www.legitreviews.com

You are stating that a 2700 in handbrake would go well over its 65w, however the above review does not support that. It does go slightly over if going from the wall power, but that is also with 3200mhz ram and a 1200w plat psu.

 

[rbk123]

Again, we don't know the detailed settings we've been talking about used in either setup. Second, is there anything different between a 2700 and a 2700X architecturally? I think not, it's just binning, but I could be wrong. Sure seems odd a 2700X can use more power but a 2700 can't...

 

[VirtualLarry]

Sure seems odd a 2700X can use more power but a 2700 can't...

As far as I am aware, that IS the case. Whatever processor-level "tunables" for TDP, PPT, etc., are set to a pretty-hard 65W TDP limit in the 2700, AFAIK, unless you do manual OC, in which case all artificial limits are removed.

 

[rbk123]

unless you do manual OC,

That's the point(s).
1. Many of the BIOS's are doing overclocks, so....
1a. You can adjust the 2700's setup to use beyond 65W,
and
2. You can adjust the 3700X BIOS settings so that it WON"T exceed 65W TDP (and it will outperform the 2700 on top of it)

 

[maddie]

Again, we don't know the detailed settings we've been talking about used in either setup. Second, is there anything different between a 2700 and a 2700X architecturally? I think not, it's just binning, but I could be wrong. Sure seems odd a 2700X can use more power but a 2700 can't...

Agreed, it's just the settings. Any attempt to portray the 2700 as a completely different beast is ridiculous. Anyone who is into building their systems can find a way to modify the behavior.

 

[Hitman928]

The 2700 by default will use more than 65 W unless you turn off CPB, same for the 3700x. The 3700x will use a little more power under heavy load with CPB on but you're talking an led light bulb type of power difference and in those situations the 3700x is giving you 40+% more performance.

 

[UsandThem]

I have to say after 4 pages of replies, this thread is something out of the show 'The Outer Limits'.

I'd never imagine that this far into 2020, that anyone would be championing a 2700 over a 3700X (at the same pricing) for really any reason. Now I would I understand if the 2700s were still like $100, but this whole thing is something you don't see very often for sure.

 

[Yeroon]

That's the point(s).
1. Many of the BIOS's are doing overclocks, so....
1a. You can adjust the 2700's setup to use beyond 65W,
and
2. You can adjust the 3700X BIOS settings so that it WON"T exceed 65W TDP (and it will outperform the 2700 on top of it)

I wouldn't call CPB as default behavior an overclock, since this is a bios setting offered and promoted by AMD, and this explains why it is initially enabled. The cpu still adheres to its base-boost frequency range.

The point a few of us were trying to make is that the 3700x can be set to stay within the 65w window, but that does require knowing more intimate details of how the 3000 series boosting works.
The problem here is finding the performance data while a 3700x is limited to its 65w 'rated' tdp, and knowing what to change in the bios to get it there. No reviews that I know do that, otherwise this would have been a very short thread.

The 2700 by default will use more than 65 W unless you turn off CPB, same for the 3700x. The 3700x will use a little more power under heavy load with CPB on but you're talking an led light bulb type of power difference and in those situations the 3700x is giving you 40+% more performance.

But yet the data available show the 2700 stays pretty close to its 65w without changing bios settings. No reviewer is going to turn off CPB only for the 2700 and then not mention it in the review.
I don't understand why this needs to be repeated every post, it is clearly visible in reviews and mentioned by the OP.

40+ watts is one hell of an LED.

I have to say after 4 pages of replies, this thread is something out of the show 'The Outer Limits'.

I'd never imagine that this far into 2020, that anyone would be championing a 2700 over a 3700X (at the same pricing) for really any reason. Now I would I understand if the 2700s were still like $100, but this whole thing is something you don't see very often for sure.

This is like reading comprehension nightmare thread.

If you know what to do, a 3700x is 100% the better purchase. I don't know who is disputing this. Being able to show this with performance data AT 65w is what is unavailable, since as all reviewers tested it, it uses far more power.
I did my best to show that a 3000 series can be made to stay within these limits.

 

[Hitman928]

But yet the data available show the 2700 stays pretty close to its 65w without changing bios settings. No reviewer is going to turn off CPB only for the 2700 and then not mention it in the review.
I don't understand why this needs to be repeated every post, it is clearly visible in reviews and mentioned by the OP.

40+ watts is one hell of an LED.

You are trying to guess what the CPU power use is by comparing system power levels between different systems with different configurations and several other unknown variables which makes your claims pointless. This has already been pointed out to you but you choose to ignore it.

It's not that complicated, we know the PPT of a 3700x, it is 88 W. That's the most it will use under heavy load with CPB enabled. It may peak ever so slightly above that in transient situations, but it won't allow a sustained power use above 88 W.

The 2700 uses 65W at base frequency of 3.2 GHz under full load. With CPB on, it will run at 3.4 GHz plus maybe a little extra depending on the motherboard. To reach the higher frequency, it will use more power. I already posted where Gamer's Nexus measured actual CPU power under heavy load and the 2700 used 75.6 W compared to 85.2 W for the 3700x, that's 9.6 W or roughly equivalent to an LED bulb. I have a 2700 and can confirm that it will use at least this much under heavy load.

As far as the 3700x performance at 65 W, I venture to say that it will be unnoticeable. Here is a test where they benchmark the 3700x under normal and eco mode settings. Even under eco mode which puts the CPU at 45 W, they see no distinguishable difference in gaming and less than a 5% performance reduction in Cinebench. Again, the 3700x will crush the 2700 when both are put to 65 W. What is there left to discuss?

 

[rbk123]

I'd never imagine that this far into 2020, that anyone would be championing a 2700 over a 3700X (at the same pricing) for really any reason. Now I would I understand if the 2700s were still like $100, but this whole thing is something you don't see very often for sure.

Exactly. Pretty sure this "new" poster had a purpose and it wan't to try and find another 2700...

 

[DrMrLordX]

Exactly. Pretty sure this "new" poster had a purpose and it wan't to try and find another 2700...

Isn't it great that this post was featured in the Anandtech weekly spam notification?

 

[UsandThem]

Isn't it great that this post was featured in the Anandtech weekly spam notification?

I didn't notice that.

Now that's funny! If I were wanting to promote more readership, this wouldn't be the one I chose as the "must read" thread of the week.

 

[rbk123]

Isn't it great that this post was featured in the Anandtech weekly spam notification?

Maybe too many posts in a period of time on this thread?

So where do I see/find the Anandtech weekly spam notification? It's rare my supreme wisdom is "featured"; might actually be a first time ever...

 

[Kenmitch]

So where do I see/find the Anandtech weekly spam notification? It's rare my supreme wisdom is "featured"; might actually be a first time ever...

Same place a PM notification shows up.

 

[thesmokingman]

Micro Center rules. I bought 4 2700x CPU's for $130 each last fall. I picked up great bundle deals on new x370 strix boards they were blowing out with them too. I've gotten a handful of amazing open box deals there over the years. I grabbed an RX 470 8 gig for $13 and a 1070 ti for $170 last year. I've picked up multiple b450 boards for less than $25 with bundles.

The store is right by my office so they tend to get a lot of my disposable income.

I'm 10 minutes from a MC. I was just there last night to pick up a waterblock and it was packed. They had the covid distancing thing going so both instore and online pickup lines wrapped all the way around to the back of the store. It was crazy. While waiting in the pickup line, I was so tempted to grab an XT cpu but they didn't have any ROG boards I'd want but they do give you 15 days to claim the $50 off an XT cpu. In case a board comes in stock in the next 15 days so you can claim the combo discount. Love MC...

 

[chrisjames61]

I have to say after 4 pages of replies, this thread is something out of the show 'The Outer Limits'.

I'd never imagine that this far into 2020, that anyone would be championing a 2700 over a 3700X (at the same pricing) for really any reason. Now I would I understand if the 2700s were still like $100, but this whole thing is something you don't see very often for sure.

"The Outer Limits" Great show btw. I have to agree. People tend to use logic and think rationally. Buying a 2700 today for the same price or more than a newer Ryzen cpu is neither rational nor logical when you think objectively. I don't think there is one positive use case anyone can come up with.