Discussion in 'Frontpage news' started by Hilbert Hagedoorn, Sep 2, 2019.
Could you maybe consolidate all your answers into one post?
To be fair to AMD, the term "Max boost clock" implies that the speed isn't guaranteed. "Boost clock", as a term, clearly states that it is dependent on many factors.
It's the same as when your ISP said your internet connection is "Up to 24Mbits", or the "Max or peak power" of your speakers.
Yes that makes them obnoxious, but the terminology used means they aren't technically lying.
But still, they did lower the speed later. They should apologize for making a mistake in the first place. It would be a blow to their image, but it's not worse than what they are actually doing now.
I avoided responding in this thread, but I will say my part.
I want a CPU that offers good value, good security and longevity of the platform.
There is very little choice between big players, Intel -AMD.
So, choosing AMD after their botched advertising about boost clocks seems like picking the lesser of two evils.
Does one company offer something for free, or cares about your bills or whatever? Let's not fool ourselves.
I'll stick to my plan of buying an 3700X at the lowest price I'll be able to find during Black Friday. Until then AGESA and motherboard manufacturers will fix some, will improve some.
And I am fully aware of this silicon lottery and special specific conditions to have your Ryzen boosting to advertised clock speeds.
Realistically, what one can do? Nobody's forcing anyone to buy something.
You choose and live with your choice.
it also implies you should occasionally see it under particular loads,
and this is clearly not the case.
I am pretty happy with my 3900X it boosts to around 4.57 which is fine and its not running nearly as hot as my I7-8700K but its running higher IDLE temps not a big issue for me. Overall i think AMD should have been more on the level with what people can expect. I get more performance in all applications and in the games i play there are some where i get the same performance and some where i get better performance, mostly my games are GPU bound by the 3440x1440 resolution.
For now, do not use the Ryzen Balanced Power Plan. Use the Windows Balanced Power Plan. If you do this, your CPU will go to the idle 2.2GHz clocks like it should, and your idle temps and idle voltage will then also be normal.
These results are non scientific, however there does seem to be some statistical evidence of course and even negating people not being able to put a hsf on properly, using different associated hardware, environment and perhaps being trolls, it would suggest an in the field difference to quoted numbers. The fact people on here say they like their AMD but do not hit the boost can also be added to the info coming out.
Of course the chip is new process, uses chiplets and is used in conjunction with 3rd party motherboards etc who also seem to be making decisions on boost levels, so it will be interesting to see where this all goes as things mature.
Some good posts above on the pro's and cons. We do have a fixation about numbers and maximums ( I'm thinking of buying a Chiron now it does 300mph, I was pretty meh before !) but it should always be about performance v cost for what you use it for. AMD really got their IPC up so it is competitive with Intel, but I think people were hoping it would overclock more on top of the claimed boost to be even more of a bargain. Instead a few AMD style teething problems giving people the opportunity to do a lot of reading into how to get the beast fully roaring
Or fully meowing if yours is not so good.
Note I am on Intel Skylake currently but owned K8 in the past, happy AMD bought out a competitive product, the market needed it.
Technically, if AMD can prove that each of those chips did actually run and pass tests at given clock, it is not an problem for them.
(Would you guess if they did binning and kept results of tests for each CPU?)
I believe they bin the ccx's themselves rather than the cpu in its entirety, as far as i can tell theres no markings on the ccx's to correlate to such results.
It's entirely possible the results are due to the benchmark used.
Boost clocks are all todo with power and thermals. Cinebench most likely uses SSE and AVX instructions which use a lot of power and generate a lot of heat. Remember, even Intel chips drop CPU frequency when AVX2 instructions are executed. It almost made those instructions pointless when they were introduced. AVX512 does the same.
A simpler more integer based workload may throw up more consistent results around the claimed boost clocks.
It's possible that AMD got out of hand with advertising with these ryzen, and if they promised one thing they should keep it.
On the other hand, the tests are there and these processors are doing well and at a good price; the R5 3600 will have to go down a bit.
Didn't "Derbauer" sell products to make "delid"?
I guess it's just me, but I have less enthusiasm to spend on a new Ryzen 3K build in the next months. I'll guess I'll postpone my build upgrade again...
I so wanted to go Ryzen, AMD, show the finger to Intel... but the pricing took a steep climb up to almost Intel level, due to my very special usage of the PC (gaming only) I can not make value out of the better general package of AMD's offering, and now boost clocks not even going there... actually less performance than with reviews in the beginning... it just feels like one has to be uncertain of what one would get when buying an new AMD rig now.
Is there any chance for a quick re-test with latest BIOS builds @Hilbert Hagedoorn ? If not I'd perfectly understand. But it feels a bit like what AMD didn't have to "security-fix" performance wise, they are losing with AGESA limits now...
That said, that's just an outside look of somebody who has neither the latest Intel CPU nor AMD CPU.
I guess I'm one of many that is actually satisfied with the chip and also for helping AMD to become that mighty rival to Intel it once was and remember, if this generation almost closed the gap on Intel, imagine what next generation will do
I think that this whole boosting problem has gone a bit out of proportion combined with a complete failure in AMD marketing. My CPU does really well in gaming which, after all, is the most important thing. Heck, you could still get that 3900X and be future proof for a long time ahead.
If I was you I'd hop on the AMD wagon now and enjoy a decent processor After all, things can only get better.
Yes it does look like a fail in marketing, and then technical marketing. Since AMD could very well have told people they're cutting down on boost clocks... such things are always painful, but even more so if people find out themselves, and not the manufacturer having announced it.
Not sure I want to jump onto the x570 Ryzen 3K bandwagon anymore... maybe even see what comes up with Ryzen 4K, slightly higher (and hopefully more consistent) clocks and without all that stuff about PCIe 4 which I wouldn't even need... I mean, they did really well just... don't need it, and I doubt it's worth the extra time tinkering with it. The older I get the less I want to bother with that kind of stuff.
Umh, no, not really as most who used to have AMD CPUs went over to Intel because they provided better performance for many years. Now, some of us are migrating back because AMD is back in the game.
If anything then it's the intel fanboys that have been OK with way overpriced CPUs (why did 'you' keep buying some of them?), and there has been very little noise given to poor IHS on Intel CPUs, or how easily they have for instance cut out hypertreading and cache just because they can, they are on the top of the hill so let them do it. And of course it's clear that they have kept the number of cores down because nobody challenges them, so they have made huge profits. We also have seen quite a bit anti-competition behavior, which most people tend to forget about.
I'm on a 3600 and about 50mhz short of the advertised max boost speed, so that's not a biggie. In the weekend I'll investigate to see the cooler sits properly as well as the thermal paste... things are so new at the moment that it's even hard to have proper readings of the temperatures.
Going to be interesting what the situation will be when I upgrade to 3950x. I've been happy with my 1800x works as advertised.
The 3900X is $1000? https://www.newegg.com/p/274-003B-00061
I think the term max boost does imply a guarantee. You should see the boost at least a little lol. A cars max speed is a real thing. I don't understand the confusion. Maybe you're just trying to be a devil's advocate but I don't think it's fair.
All that is possible
Out of stock in any shop here and on amazon.de one new is already 800 euro...
Intel i9900k craziness all over again but with a different company...
Zen and Zen+ also had "max boost clock" on their product pages - there was no outcry about that because the processors actually hit those frequencies regularly. It's hard to fault people for expecting the same to be true for Zen 2.
I've had this 3900x since launch, under water cooling, on two different motherboards across a range of BIOS's - it's never hit it's max boost clock ever.
People also keep ignoring that newer BIOS's have since lowered the maximum clock speeds for multiple people (me included), ASUS reported that AMD did this for reliability reasons. They also ignore that AMD claimed in the PBO video, prior to Zen launching, that potentially higher than max boost was possible given x, y and z conditions. In the video they talked about potentially hitting 4.7ghz with the "right cooling and power" - in the mean time I can't even hit what was advertised to me with both those requirements easily checked off.
Is the lack of frequency disappointing? Not really - processor still performs extremely well but this type of behavior needs to be called out. You can't let companies get away with this crap or it becomes the norm. Soon they'll be shipping review processors that boost to 5Ghz but the retail units will be changed to 4.4 and people will still sit here and be like "it's fine because it says "up to"" or some bullshit.