Overclock.net banner

121 - 140 of 254 Posts

·
PC Evangelist
Joined
·
47,375 Posts
Because AMD needs to care about OCN when they release a new product.
Not at all. I like it better this way. Extract all the performance already out of the box since 95% will use it that way. All I am trying to say is why all these beefed up MB with crazy VRM cooling and designs for a CPU that cant really OC. You want these MB for X299,X399 but 105W TDP?

Also all the talk about Ryzen being "Unlocked". To me, this is like Fury X all over again. Ryzen = FX95XX mentally coming from AMD. This would be fine if 2700 non X would be worth the buy but for $30 and better cooler, u get 2700X and no need to bother overclocking.
 

·
Registered
Joined
·
828 Posts
http://hardwarebg.com/44332-ryzen-7-2700x-ryzen-5-2600x/6/ Has anyone seen this yet?

Basically there is rumblings, that Intel asked the reviewers to benchmark with HPET off in the BIOS, and it has drastically skewed results in favour of the 8600K/8700K.

The site linked shows bench results with and without HPET enabled in bios for the Intel chips.

Pretty big if its true, and could explain why Anandtech's results are such an outlier.

Intel? engage in underhanded behavior to manipulate the market in their favor? Well, I never!
 

·
Smug, Jaded, Enervated.
Joined
·
1,278 Posts
Intel? engage in underhanded behavior to manipulate the market in their favor? Well, I never!
Indeed! How unprecedented!!!
 

·
Registered
Joined
·
3,230 Posts
Not at all. I like it better this way. Extract all the performance already out of the box since 95% will use it that way. All I am trying to say is why all these beefed up MB with crazy VRM cooling and designs for a CPU that cant really OC. You want these MB for X299,X399 but 105W TDP?

Also all the talk about Ryzen being "Unlocked". To me, this is like Fury X all over again. Ryzen = FX95XX mentally coming from AMD. This would be fine if 2700 non X would be worth the buy but for $30 and better cooler, u get 2700X and no need to bother overclocking.
I agree. A lot of people are going to get the 2700x just for the cooler. I wonder if the average person would really pawn off the prism? It's so cool. This is coming from a person that doesn't really like rgb that much. But anyways,yeah just buy 2700x. Enjoy the rgb cooler and that's it. The turbo boost algorithm pretty much takes care of you.

1ghz+ oc is dead. This is the core wars.
 

·
sudo apt install sl
Joined
·
8,338 Posts
http://hardwarebg.com/44332-ryzen-7-2700x-ryzen-5-2600x/6/ Has anyone seen this yet?

Basically there is rumblings, that Intel asked the reviewers to benchmark with HPET off in the BIOS, and it has drastically skewed results in favour of the 8600K/8700K.

The site linked shows bench results with and without HPET enabled in bios for the Intel chips.

Pretty big if its true, and could explain why Anandtech's results are such an outlier.
Doubt that's the reason why since the 8700k FPS is similar to the 8700K release review anandtech did. If anything the scores went up a few frames due to maybe Graphic drivers? Some how they butchered their 2700x setup.

https://www.anandtech.com/show/12625/amd-second-generation-ryzen-7-2700x-2700-ryzen-5-2600x-2600/16

https://www.anandtech.com/show/1185...lake-review-8700k-and-8400-initial-numbers/15
 

·
Sunday League Jibber
Joined
·
4,254 Posts
They are, I'm disappointed by AMD that even with Intel not releasing any IPC improvements with the past two generations their chips are still slower.
It's almost as though they had to recover from a disastrous attempt at redefining the trajectory of personal computing and do so on a shoestring budget relative to their primary competitor. Nevermind that AMD is now competitive in almost every scenario bar 1080p VHRR.
 

·
Not a linux lobbyist
Joined
·
2,455 Posts
You can benchmark under artificial conditions if you'd like to, but since none of us play at 640x480 with all settings set to minimum it doesn't do anyone much good to know how processors perform under such conditions. It might theoretically show the difference between two different platforms, but it wouldn't be any more meaningful to someone seeking information about real world performance than would harping on cache access times. It means something, but it doesn't tell anyone how their game is going to run under the conditions they're likely to play them.
While it is true that most will opt to push their graphical limits first, and for me at [email protected] a Ryzen would bottleneck my 1080tis, it is dishonest to imply that a cpu can put out higher fps when the limiting factors lie elsewhere. A lot of people want more than they need. But yes, a Ryzen or hyperthreaded quad core haswell or sandy bridge is still fine for most normal people.
And as long as you are talking about real world, how many are actually ever going to install the spectre resistant bios? Or not overclock an unlocked intel cpu?

Edit: and that 720p testing was for seeing the difference between the spectre microcode and the others, not to compare against some cpu I don't have.
 

·
Sunday League Jibber
Joined
·
4,254 Posts
While it is true that most will opt to push their graphical limits first, and for me at [email protected] a Ryzen would bottleneck my 1080tis, it is dishonest to imply that a cpu can put out higher fps when the limiting factors lie elsewhere. A lot of people want more than they need. But yes, a Ryzen or hyperthreaded quad core haswell or sandy bridge is still fine for most normal people.
And as long as you are talking about real world, how many are actually ever going to install the spectre resistant bios? Or not overclock an unlocked intel cpu?
Anecdotally, a lot. In my gaming circle I know 30+ people with -k processors and only four of them are overclocking them. The mass market sees the higher stock clock and buys the -k chip.

I'm not against testing pure CPU bottlenecks for the edification of hardware nerds, but I agree with Particle that such tests don't accurately reflect the majority of real-world scenarios. If Anand prefers to discuss real-world differences, that's fine. The Stilt writes exhaustive reviews in their forums to satisfy even the most abstract of theorists.
 

·
PC Evangelist
Joined
·
47,375 Posts
I agree. A lot of people are going to get the 2700x just for the cooler. I wonder if the average person would really pawn off the prism? It's so cool. This is coming from a person that doesn't really like rgb that much. But anyways,yeah just buy 2700x. Enjoy the rgb cooler and that's it. The turbo boost algorithm pretty much takes care of you.

1ghz+ oc is dead. This is the core wars.
Not even asking for 1GHz. Cant we just have the ability to OC to a least a bit over rated Boost clocks? It seems if you OC 2700X u lose performance unless you can do 4.3GHz+ in all cores.
 

·
Not a linux lobbyist
Joined
·
2,455 Posts
Anecdotally, a lot. In my gaming circle I know 30+ people with -k processors and only four of them are overclocking them. The mass market sees the higher stock clock and buys the -k chip.

I'm not against testing pure CPU bottlenecks for the edification of hardware nerds, but I agree with Particle that such tests don't accurately reflect the majority of real-world scenarios. If Anand prefers to discuss real-world differences, that's fine. The Stilt writes exhaustive reviews in their forums to satisfy even the most abstract of theorists.
I only know 7 with -k processors and all overclock but not all increase voltage and I'm the only with a delid. But it is mostly a blue collar crowd in WI and not representative of society as a whole, so that is probably where I got that impression. You know of enough that don't so I guess a lot don't.

If you don't oc, this is a good cpu. As good for most gamers/ office users and web browsers as the 4790k at least. Certainly good enough to gpu limit most.

That spectre resistant microcode is going to make a real mess of relative benchmarks. It is hard to even get it on some pre skylake mobos.
 

·
sudo apt install sl
Joined
·
8,338 Posts
I agree. A lot of people are going to get the 2700x just for the cooler. I wonder if the average person would really pawn off the prism? It's so cool. This is coming from a person that doesn't really like rgb that much. But anyways,yeah just buy 2700x. Enjoy the rgb cooler and that's it. The turbo boost algorithm pretty much takes care of you.

1ghz+ oc is dead. This is the core wars.
Not sure what you mean by 1ghz+ oc is dead when Intel's 7820x are hitting 4.6 which is a 1ghz overclock. The 2700x is overclocking by 600mhz which isn't terrible but it's on a new build process. Intel's been on 14nm for 4 generations and refining it.
 

·
Top kek
Joined
·
3,595 Posts
http://hardwarebg.com/44332-ryzen-7-2700x-ryzen-5-2600x/6/ Has anyone seen this yet?

Basically there is rumblings, that Intel asked the reviewers to benchmark with HPET off in the BIOS, and it has drastically skewed results in favour of the 8600K/8700K.

The site linked shows bench results with and without HPET enabled in bios for the Intel chips.

Pretty big if its true, and could explain why Anandtech's results are such an outlier.
First time i see a review from my own country, where it actually finds something good. I usually dont bother reading at this forum, but damn, kudos to them.
 

·
⤷ αC
Joined
·
11,239 Posts
Ok, so I know everyone seems to like Ryzen but I have 2 problems with it.

1) Its a middle finger to OCN.
2) Why spend money on expensive X470 MB for the above reason?
Did you ever stop to reevaluate your perspective? Maybe change the Intel-skewed lens you have and look at it from an unbiased perspective?

If you have been paying attention some of these early reviews are flawed. Some don't have Spectre + Meltdown patches for the Intel chips. Most aren't locking apps to a CCX for lightly threaded workloads (<4 cores or <8 thread) via Process Lasso or CPU affinity. There's no real large reason for the Ryzen 2000 series chips to be outperformed by anything locked Intel i5 8600 in non AVX2 workloads and the performance differential should be below 10% vs a locked i7-8700 if only ~4 threads are used.

Some people have been achieving 4.45GHz overclocks but with high voltage along the lines of 1.5V. 4.2GHz overclocks are "quick and dirty" all core clocks and the 12nm leading performance process used has more leakage so there's more variance. You have to keep in mind a 4 core has much higher probability to have a certain clock since there's only (1/2)^4 probability of hitting a clockspeed versus (1/2)^8. Obviously the events are not truly independent since the location of the die on the wafer factors into every core.

As far as stock performance I'm not quite sure reviewers were making full use of Precision Boost Overdrive.
The Stilt's "Performance Enhancer option can increase or disable XFR power and current limits allowing you to boost higher and longer. Together with reference clock adjustments this means you can get up to ~4.5 GHz in single threaded loads and 4.2-4.3 GHz in multi threaded." --- CH VII Hero thread

The reason for the X470 lineup is the VRM improvements to memory and also the B350 and some X370 boards have anemic power delivery.


https://forums.anandtech.com/threads/ryzen-strictly-technical.2500572/page-72 said:
The "Precision Boost Override" feature available on 400-series motherboards allows increasing the physical limiters mentioned earlier. On SKUs belonging to the 105W TDP infrastructure group, the default limiters are following: PPT 141.75W, TDC 95A, EDC 140A and tJMax of 85°C (absolute, excl. offset).

When "Precision Boost Override" mode is enabled (AGESA default), PPT becomes essentially unrestricted (1000W), TDC is set to 114A and EDC to 168A. These limits can be customized by the ODM so that the new limits will comply with the electrical characteristics of the motherboard design in question.
Package Power Tracking (PPT) , Thermal Design Current (TDC) , Electrical Design Current (EDC) are all factored in now. Per CCX overclocking also is involved as well as fastest core detection (for all Ryzen CPUs using the Ryzen Master Utility).

Per the Stilt, the out of the box settings for these constraints are as follows:
PPT 141.75W, TDC 95A, EDC 140A and tJMax of 85°C

Most boards on B350 as well as some X370 boards wouldn't be able to safely handle this (mainly the 140A EDC). The top tier X370 boards such as the CH VI Extreme / Hero , Taichi (same VRM for X470 version), X370-F STRIX (same VRM as X470-F STRIX), Biostar GT7 should be able to. People that were foolish to buy a $80 B350 board for a longterm AM4 socket aren't going to be able to make full use of a R7 2700X.

168A is well beyond what I would attempt on a X370 midrange board cooled by an indirect case fan since most are not properly doubled and just 8 so-called phases on 4 interleaved PWM signals.
 

·
PC Evangelist
Joined
·
47,375 Posts
Did you ever stop to reevaluate your perspective? Maybe change the Intel-skewed lens you have and look at it from an unbiased perspective?

If you have been paying attention some of these early reviews are flawed. Some don't have Spectre + Meltdown patches for the Intel chips. Most aren't locking apps to a CCX for lightly threaded workloads (<4 cores or <8 thread) via Process Lasso or CPU affinity. There's no real large reason for the Ryzen 2000 series chips to be outperformed by anything locked Intel i5 8600 in non AVX2 workloads and the performance differential should be below 10% vs a locked i7-8700 if only ~4 threads are used.

Some people have been achieving 4.45GHz overclocks but with high voltage along the lines of 1.5V. 4.2GHz overclocks are "quick and dirty" all core clocks and the 12nm leading performance process used has more leakage so there's more variance. You have to keep in mind a 4 core has much higher probability to have a certain clock since there's only (1/2)^4 probability of hitting a clockspeed versus (1/2)^8. Obviously the events are not truly independent since the location of the die on the wafer factors into every core.

As far as stock performance I'm not quite sure reviewers were making full use of Precision Boost Overdrive.
The Stilt's "Performance Enhancer option can increase or disable XFR power and current limits allowing you to boost higher and longer. Together with reference clock adjustments this means you can get up to ~4.5 GHz in single threaded loads and 4.2-4.3 GHz in multi threaded." --- CH VII Hero thread

The reason for the X470 lineup is the VRM improvements to memory and also the B350 and some X370 boards have anemic power delivery.




Package Power Tracking (PPT) , Thermal Design Current (TDC) , Electrical Design Current (EDC) are all factored in now. Per CCX overclocking also is involved as well as fastest core detection (for all Ryzen CPUs using the Ryzen Master Utility).

Per the Stilt, the out of the box settings for these constraints are as follows:
PPT 141.75W, TDC 95A, EDC 140A and tJMax of 85°C

Most boards on B350 as well as some X370 boards wouldn't be able to safely handle this (mainly the 140A EDC). The top tier X370 boards such as the CH VI Extreme / Hero , Taichi (same VRM for X470 version), X370-F STRIX (same VRM as X470-F STRIX), Biostar GT7 should be able to. People that were foolish to buy a $80 B350 board for a longterm AM4 socket aren't going to be able to make full use of a R7 2700X.

168A is well beyond what I would attempt on a X370 midrange board cooled by an indirect case fan since most are not properly doubled and just 8 so-called phases on 4 interleaved PWM signals.
Nothing in my post talks about performance.
 

·
Premium Member
Joined
·
749 Posts
I ran home at lunch because my 2700x was delivered from Newegg.

I swapped out my OC'd 1800x (4GHZ on all 8 cores) for a stock 2700x. Ram (2-8gig sticks of GSkill Flare-X 3200) was kept at 3200 speed and the BIOS for my Asus Crosshair VI Hero is the newly released 6004.

All went well. On the 2 synthetic benchmarks I had time to run before I had to go back to work (Aida64 Queen and Cinebench R15) the stock 2700x scored slightly higher than my OC'd 1800x.

Ex. Aida64 Queen 1800x 92300 2700x 93300
Cinebench R15 1800x 1760 2700x 1792
 

·
Registered
Joined
·
3,230 Posts
Not sure what you mean by 1ghz+ oc is dead when Intel's 7820x are hitting 4.6 which is a 1ghz overclock. The 2700x is overclocking by 600mhz which isn't terrible but it's on a new build process. Intel's been on 14nm for 4 generations and refining it.
The 2700x boosts itself over 3.9ghz-4ghz out of the box. Maybe even higher with XFR2 and depending how many cores are being used. When one core is being used it should boost to 4.3ghz and it should boost to 4.35ghz when XFR2 kicks in. But that's only going to be for a split second. On and off. On and off. If temps are good and what not. So if you're overclocking and hit a wall at 4.2ghz you lose XFR. You are fixed at 4.2ghz on all cores. Unless you get into p-state overclocking.

There's a lot a different ways of looking at it. And yes the 7820x can hit 4.6ghz. 8700K can 5.3ghz. I said can. Doesn't mean everyone. :)
 

·
Optics Fiend
Joined
·
4,856 Posts
Not even asking for 1GHz. Cant we just have the ability to OC to a least a bit over rated Boost clocks? It seems if you OC 2700X u lose performance unless you can do 4.3GHz+ in all cores.
Least i was hoping for 4.3 and it looks like we've more or less got it with adequate cooling. I mean what's the metric here? 4.2ghz Ryzen is within 10% of a 4.7ghz or so Coffee Lake in gaming which is really the one Achilles heel that has plagued Ryzen. Not to mention $30 cheaper, 2 more cores, and comes with a heatsink?

Don't get me wrong i see the point i wan't higher clocks, but just seems like we'll have to wait for Ryzen 2 to get there.
 

·
Debian Dude
Joined
·
2,184 Posts
it is dishonest to imply that a cpu can put out higher fps when the limiting factors lie elsewhere
It's not a matter of interpretive honesty. Test the two different processors under normal conditions and report the results. They'll probably be the same, but the point of testing is to see what really happens. Theory isn't as valuable as experimental data.

Dishonest would describe the practice of testing a game under conditions nobody is going to play at.
 

·
Not a linux lobbyist
Joined
·
2,455 Posts
It's not a matter of interpretive honesty. Test the two different processors under normal conditions and report the results. They'll probably be the same, but the point of testing is to see what really happens. Theory isn't as valuable as experimental data.

Dishonest would describe the practice of testing a game under conditions nobody is going to play at.
If you like gaming 120-144hz and are ok with sacrificing some graphical settings to get it, and you get a Ryzen under the impression that it can handle that better than a 5ghz 8700k, and it can't (it might, but Anand's review only tests 2 of 5 games with that fps range, so you don't know, but are given a clear impression that it should beat the 8700k at everything) then it is a matter of honesty, not interpretive honesty.

I don't think most people know that an 8700k can put out far more fps than they are showing with either a more powerful graphics card or reduced settings. Can a 2700x? Probably somewhat better than the 1700, but this article obfuscates that.
 
121 - 140 of 254 Posts
Top