Read more.Quote:
For example, only 5.6 per cent of Ryzen 3900X users could boost to the advertised clock speeds.
Printable View
Read more.Quote:
For example, only 5.6 per cent of Ryzen 3900X users could boost to the advertised clock speeds.
Is that a 'class action' law suit I see approaching on the horizon?
Sadly you might be right. I always see way more than base clock even when hammering all threads for some minutes on the retail cooler, so I actually think AMD have been quite generous with their ratings for where it really matters. That won't stop the vultures from circling at the first opportunity though.
Personally I would like to see a controlled test rather than random users with different hardware taking random times to do the test.
I'm surprised Hexus published this as a "news" article. Should have just said bloke gets random users with random hardware to do a test. Results inconclusive.
I am quite shocked that people are not able to open an image, think and then read what the click-baity article says. As you can see, the majority of samples for 3900X boost to 4525Mhz, which is 75Mhz lower than advertised boost speed. If that isn't close enough I don't know what it is...
Also, to prove that the title is click bait, it should have been: "Ryzen 3000 boost is very close to the one advertised in most cases". No, they transformed this 75Mhz gap which is 1-2% into a huge deal breaker and they seem to be insisting with this since there are not many bad things to say about the new AMD cpus...
Matter of fact is that these new CPUs are great performers, great prices, great consumption and I sense a bit of hatred from Intel supporters since they don't have many things to say. Also lets not forget about that Israeli firm that made a site and some videos specially for AMD to denigrate them. It is up to consumers to use their brains and not be fooled.
Also, most motherboards will only run "close to" the advertised speed. I routinely see intel chips running up to ~25Mhz lower than quoted speeds just because of how the clocks run and how the BIOS is configured. Effectively there's a small margin in the various clockings that determine CPU speed, and mobo vendors always err on the side of caution to avoid rma issues.
This is probably different, due to the magnitude of the discrepencies (75 Mhz is a much bigger difference), but so many factors affect this that it is hard to be certain when it is a "boost clock". One odd thing, though, is that I thought precision boost (or similar name) was supposed to push *beyond* stock clocks if there was suitable headroom, so I'd have expected slight overclocks not underclocks on 3x00X parts?
My 3700x is a strange beast all core of 4ghz and single core of 4.3ghz.
If I enable PBO I get all core boost of 4.16ghz and single core of 4.2ghz.
Always running rather cool with a Nepton 140XL.
I wouldn't exactly call it an uncontrolled test as the parameters were defined and isn't having random users the entire point, we're not talking about what the max boost is in strictly controlled laboratory conditions here, we're talking about what your average customer would achieve out of the box, IDK what times have to do with it or different hardware as it was the maximum boost speed recorded that mattered, if different hardware effects that then shouldn't AMD make people aware that they should be using X hardware.
I think they can easily say 25mhz-75mhz is within a culpable margin of error.
Frankly, this is better than Intel advertising 5GHz boost on U/H processors of which the PL1 value only lasts for less than a minute or so before thermal dial back to PL2/3 to base clock or even under base clock. That is far more disingenuous.
Steve from hardware unboxed this a scientific test with 1 CPU and found that based on different motherboards it performed differently. On some it wasn't getting the proper boost, on others it was.
I think it depends on motherboard, cpu, cooling, OS, bios version, chipset version, and which program you are running, therefore everyone is seeing different boost speeds. Plus who the idiots out there, boost clocks are opportunistic clocks, amd does not guarantee them, they only guarantee base clocks, so any notion of a lawsuit is absurd.
Got to be honest I don't take much notice of boost speeds as my main interest is what it's like with all cores... I fully expect that any amd ryzen 3 cpu I buy will be able to run at the base clock speed on all cores 24/7, just like I expect my i7 4790k to be able to run at it's stock 4ghz on all cores 24/7...
Boost clocks to me have so many variables, and heat being a major factor. The UK, as did most of EU iirc, just had a heatwave and a I saw 10-20 degrees higher temps under load and I'm sure that would have actually impacted some of those results seeing as they're from the general public.
Should AMD have been so aggressive in their 'max boost clocks' assertions, probably not but I remember a time when boost clocks weren't even a thing...at the same time I don't think it's a class action worthy issue either because of the amount of variables at play AND the fact that AMD have used max boost clocks rather than saying straight up it will clock to this speed.
Have to agree LSG501, I always looked at boost speeds as a free overclock, never got upset because I couldn't achieve them, as long as it run it's base speed was happy
The hardware aspect of pc's for gaming is full of guys talking amd vs intel overclocking custom water loops etc best air coolers. The game side of PC market is full of sjw marketing #metoo making baseless allegations while game devs hang themselves.
Let's be honest. If this were intel, they would get so much aggro. I don't regret my 3600, but AMD did seem somewhat misleading about the boosting potential.
Dunno what all the fuss is about as I have a 2600 ;)
I doubt that. Where was the uproar about Intel's PL2 power states? Quite.
AMD were right to quote a max boost clock as long as retailers pass on that important 'max' modifier. I do want to know what the maximum turbo frequency is, and I want to know what the base clock is. If you always expect to see a turbo frequency then that's the base clock, not the max boost frequency.
https://twitter.com/AMDRyzen/status/1168901636162539536
https://pbs.twimg.com/media/EDjF1zmX...pg&name=medium
I think the trouble with that is that users of high end CPUs like this have now come to expect their CPU to reach the maximum boost frequency quite often, perhaps not for prolonged periods but at least some of the time.
I agree though, I'd rather see AMD giving the CPU some genuine shtick, pushing it as hard as they can, even if it will only reach those maximum clock speeds very rarely.
UPDATE:
THERES A BIOS/FIRMWARE BUG
https://twitter.com/AMDRyzen/status/1168901636162539536
https://pbs.twimg.com/media/EDjF1zmX...pg&name=medium
(what hoonigan said, but said again as it's worth mentioning twice :P)
If that's the case then AMD should've informed customers that that's the case so they can make an informed decision.
EDIT: Posted that before i read ^^that^^ announcement, better late than never as this boost issue has been being discussed since 1.0.0.3 AGEAS and no one could nail it down, they could've save people a lot of time there. :)
I'm surprised this nonsense is getting traction in the tech press...Well, maybe not so surprised.
Boost speeds will always depend on M/B. BIOS. Cooling solutions etc.
No-one in the Tech Press seems to mind Intel quoting TDP figures that bear no resemblance to the real world.
Go figure....
Now you just have to beat the mobo makers to you know... update their rubbishrubbishrubbishrubbishty bios's...
Whats that got to do with anything? Also I never said random guy so where you getting that from?
meaning he understands the hardware and has been seeing "odd" results. So getting the community to crowdsource data is a smart move. And now AMD has confirmed issues. Problem is that now you have to wait for bios updates (if they come).
Theres been plenty of issues on bios forums as ppl try to dial in their new chips. Its still bit of a mess depending what board you have. Also now you have to be careful updating if you have older chips as what was stable gets narfed in the new patch fest.
If within 1% of advertised frequencies is a mess, i wonder what adjectives you would use to describe 2% variances.
I buy my hardware based on the performance, not on what the box says. A lot of people are still obsessed with the raw speed rather than performance/clock (or more importantly performance/dollar) and I think that's why they've risked putting a number they knew the parts couldn't quite reach.
Although I do think we should take them to task, because slippery slope and all that, I think it's important for the reaction to be measured
The issue is/was that the performance was measured using AGESA 1.0.0.2 and (afaik) AMD updated review guides 2-3 days before release to say reviewers should us the new 1.0.0.3 AGESA, obviously very few reviewers had time to retest everything so the published performance figures on release where using the un-buggy version, people who bought the product however ended up using the buggy version (at least if they updated their BIOS they did) and so did not get the performance reviewers got.
Either way it seems they're talking about a September the 10th release for the update so depending on how much testing board partners do it should be addressed shortly after that.
I think anything we do that makes chip makers timid about how aggressively their devices can boost would be bad.
If AMD were forced here to make sure that 100% of their chips can always get the full boost frequency on the cheapest motherboard in a badly vented case then they may feel compelled to cripple the boost tables for all chips by 200MHz and advertise that figure instead. I like that my chip in my system will always try and do the best it can.
If you ever see "maximum" or "up to" you know full well real world conditions will not provide that.
I thought it was well understood that the AMD chips are essentially self overclocking to the maximum that bit of silly-cone can achieve and that, like all overclocking, depends on many factors outside the CPU.
I also thought it was well understood that to make maximum use of this facility, you needed aftermarket (ideally water) cooling.
I don't see 75MHz from randomly built, configured and operated PCs as being an issue. I would see it as an issue if a properly built, optimised and configured PC could not achieve the stated boost clocks.
This is like putting that silly petrol in your car that says "up to 30% more mileage!" and then complaining it makes only 10% difference in the real world. The advertised maximum is just that.
Intel saying you can do 5GHz on all cores and having a small fridge worth around a grand plumbed into the CPU is clearly an issue with disingenuous marketing.
I hope they have found a BIOS issue that will shut this up because it's blown out of all proportion.
Except that when you overclock/over-volt or whatever, AMD doesn't guarantee ANYTHING! Max boost is opportunistic, as has been emphasized ad nauseum deoending on whatever thermal headroom might be available and depending on whatever mobo/firmware etc.
So convenient for the whiners and moaners like you to neglect to mention that!
If it's opportunistic depending on thermal headroom then they got something wrong as in a GN video they showed a 3900X only reached its advertised boost clock when it reached -80c.
Not that it matters much now as going on tests conducted on a leaked update it seem they've fixed things.