You don't need to convince me, im convinced already
Lower? It's topping the entire chart, what are you talking about?
- Aug 8, 2024
- 10
- 18
- 15
I dont think so, maybe for comparison with ryzen 7000 series it has correlations, as we see with 9700X and 7700 (not X), but I dont think that it will be correct for intel CPUs due to their lagging chip technology.
I can assure you, the 14900k at 88w will be much faster and more efficient than the 9700x. That's why YCCC refuses to run it 😎
- Aug 8, 2024
- 10
- 18
- 15
why there are no power restrictions for ryzen CPUs? It will be much more interest chart.You don't need to convince me, im convinced already
Lower? It's topping the entire chart, what are you talking about?
Do you realize you literally linked a chart that shows 11 Zen CPU's topping in efficiency before the first Intel shows up? You are trying to argue efficincy and that is the worst argument you can try to make. Also, doesn't seem your sig has aged well. I'm to lazy to check it out because I'm sure you will come back with some kind of pretzel logic and it isn't worth my time to prove someone on the Internet wrong. Something you seem to be very emotionally invest in.
Out of the box settings do not equate efficiency chief. If that was the case then the 7950x would be less efficient than the 5950x, but it isn't, it just has a higher power limit.
It aged pretty great actually. TBF, I was expecting the 9700x to be slightly faster than the 12700k, instead it was slower.
- Aug 8, 2024
- 10
- 18
- 15
Maybe he right, intel CPUs has power effective cores and if we restrict whole power consumption for CPU (etc for productive cores) - the intel power effective cores makes the difference? Ofc we can deactivate that cores and test again 😀
Yup because people will pay top notch for the highest end Intel CPU and limit its perfromance drastically by lowering wattage rathen than buying a cheaper model that performs the same. /s
you could run yours right? I won't run it at default VID, I REFUSE to trust intel without undervolting, and if undervolted, where is the fairness? Or you can buy me a 7950X setup and i guarantee I will test the 14900k into oblivion
you know you are comparing a TUNED bloody efficiency compared to stock competition? and competition cana be tuned also, so the point being? not to say it is really a genius move to buy a TOTL $600 part to get all those cores just to under power it and compete with something with half or even less than half the threads, it makes perfect senseOut of the box settings do not equate efficiency chief. If that was the case then the 7950x would be less efficient than the 5950x, but it isn't, it just has a higher power limit.
It aged pretty great actually. TBF, I was expecting the 9700x to be slightly faster than the 12700k, instead it was slower.
No, they won't, if they don't care about efficiency. If they do care about it, yes, why not? A cheaper model might perform similarly - but it will draw a lot more power. The same applies to both amd and intel. The upcoming 9950x, restricted to 88w will absolutely annihilate the 9700x in both performance and efficiency, so where is the problem with restricting it? I don't get it.
What tuned part am i comparing? The graph is running stock CPUs, what are you talking about?
lower the power limit or undervolting is a tuning, inteld never produce a well binned 8P16E part and make it a 65W power draw CPU, you can also limit the power of the ryzen and they may/maynot be more efficient.
Ever wonder why only intel got all those power limitation tests? it's because they need to shoot for crazy power consumption to retain performance lead in some/most situations! if they are of same TDP who would bother power limit a 95W part?
But my graphs had a stock 12700k beating a stock 9700x, what power limit are you referring to?
Saying "you can power limit the ryzen part" doesn't make sense. If it's already at 88w, and then you limit a different CPU also to 88w to compare, why would you power limit again? I don't get what you mean by that.
The best way to measure efficiency is at ISO power. That applies to pretty much everything. Like with fans, you know - you limit them to the same DBA and then you test for air flow? With cars, you don't measure mpg by having one car running 50km/h and the other one running 200km/h. You do it at ISO.
stuff and nonesense
Reputable
- Mar 10, 2020
- 508
- 425
- 5,270
Power requirements are based on what is needed to keep the thing working at spec.
Consider a cpu, it’s a huge current sink, 88W at 1V (nominal value for voltage) is an 88A draw. Many tiny transistors switching up to 5.4 billion times a second requires precise timing and granular power control.
Efficiency and power reduction
As the power is decreased the ability of a transistor to get to logic 1 is diminished. There is a rise time, logic 1/rise time gives you the slew rate. The slew rate for a particular chip can be found on its datasheet. With regard to CPUs the slew rate for a block of logic is accounted for at design time. Given a known rise time for the transistors on the silicon in those CPUs a supply voltage range is specified with tolerances for minima and to stop it burning up, maxima. Outside of the tolerances there is a grey area where the chips will still function but the performance may be degraded/unstable in the case of low voltage or simply head towards burning out in the case of high voltages.
If you reduce power
Power is volts x amps or power is amps squared x r
where r is the cpu resistance
Reducing power reduces the ability of the processor to maintain the slew rate, as the current draw is restricted by reducing the power then at a point the voltage supply, the push, will reduce as you are trying to draw more than the psu (vrm) is configured to supply.
If you reduce power within the design limits then yes, you can be more efficient to some extent. Reducing power can restrict the speed to which the cpu will clock. Good silicon, good transistors will switch to a higher speed than bad so good silicon at lower power will maintain a high(er) clock rate. Poorer quality silicon will fail, it won’t be able to switch fast enough (slew rate).
Overclocking without increasing power pushes the CPU into the “low power” state, the transistors cannot switch to logic 1 in time and a fault is seen. The solution is to increase the power giving a bigger voltage push. The drawback is that the switching can create a ringing which must settle, it’s all a balance, slew rate + ringing = settling time. The ringing is exacerbated by higher voltages.
Efficiency by reducing power supply can be enhanced to a point, that being the least good transistor failing, falling out of tolerance and not switching in time. The specs published for a device have a tolerance, psu A != psu B
If you stay within the tolerances chances are that you will be fine in normal use, under volted and under powered but when you update your bios, reset to defaults before doing so.
Consider a cpu, it’s a huge current sink, 88W at 1V (nominal value for voltage) is an 88A draw. Many tiny transistors switching up to 5.4 billion times a second requires precise timing and granular power control.
Efficiency and power reduction
As the power is decreased the ability of a transistor to get to logic 1 is diminished. There is a rise time, logic 1/rise time gives you the slew rate. The slew rate for a particular chip can be found on its datasheet. With regard to CPUs the slew rate for a block of logic is accounted for at design time. Given a known rise time for the transistors on the silicon in those CPUs a supply voltage range is specified with tolerances for minima and to stop it burning up, maxima. Outside of the tolerances there is a grey area where the chips will still function but the performance may be degraded/unstable in the case of low voltage or simply head towards burning out in the case of high voltages.
If you reduce power
Power is volts x amps or power is amps squared x r
where r is the cpu resistance
Reducing power reduces the ability of the processor to maintain the slew rate, as the current draw is restricted by reducing the power then at a point the voltage supply, the push, will reduce as you are trying to draw more than the psu (vrm) is configured to supply.
If you reduce power within the design limits then yes, you can be more efficient to some extent. Reducing power can restrict the speed to which the cpu will clock. Good silicon, good transistors will switch to a higher speed than bad so good silicon at lower power will maintain a high(er) clock rate. Poorer quality silicon will fail, it won’t be able to switch fast enough (slew rate).
Overclocking without increasing power pushes the CPU into the “low power” state, the transistors cannot switch to logic 1 in time and a fault is seen. The solution is to increase the power giving a bigger voltage push. The drawback is that the switching can create a ringing which must settle, it’s all a balance, slew rate + ringing = settling time. The ringing is exacerbated by higher voltages.
Efficiency by reducing power supply can be enhanced to a point, that being the least good transistor failing, falling out of tolerance and not switching in time. The specs published for a device have a tolerance, psu A != psu B
If you stay within the tolerances chances are that you will be fine in normal use, under volted and under powered but when you update your bios, reset to defaults before doing so.
Ah AMD..
Aging like fine wine because they release it mid fermentation and we get to wait years for the true product.
From what i can tell comparing reviews from DeBauer, Lvl1techs, GN, HUB etc... it really matters which motherboard you use, which bios is on it, and how you tune the memory.
I suspect AMD will release updated bios's between now and the 9950x release that will fix a lot of this, but it seems we really are back in the woods of the early 2000s when playing with bios settings could have huge dividends on cpu performance.
.... not sure if that's good nor not....
for the average user, definitely not.
Aging like fine wine because they release it mid fermentation and we get to wait years for the true product.
From what i can tell comparing reviews from DeBauer, Lvl1techs, GN, HUB etc... it really matters which motherboard you use, which bios is on it, and how you tune the memory.
I suspect AMD will release updated bios's between now and the 9950x release that will fix a lot of this, but it seems we really are back in the woods of the early 2000s when playing with bios settings could have huge dividends on cpu performance.
.... not sure if that's good nor not....
for the average user, definitely not.
in tech power up chart, 12700k multithread efficiency is lower than 7700X, and 9700x is more efficient, so... I don't get the logic
you tune a part to lower it's power limit to get better efficiency, coz the stock setting is not efficient, then don't allow the competitor product, costing half, having less than half the cores to down it's power limit to attain the best efficiency to it's power/efficiency curve, that is... where is the goal chief?
I did not compare their efficiency, guy questioned my sig. But efficiency between them is similar at same power.
But if you wanna test at ISO power what's the point of lowering the power lower than the target you test for? I don't get what you mean man. I really don't.
Also which competing product is cheaper? The 13700k is currently cheaper and beat's the 9700s socks off.
At 88w it's faster than the 9700x is at 170, lol.
It could be, but that's fine, aging like fine wine will get you progressively more power than when you decided it bites your price point, aging like fine milk is... disasterous
stuff and nonesense
Reputable
- Mar 10, 2020
- 508
- 425
- 5,270
Ah, but they do, gamers are a rounding error.
Corporates are the target, if they can save money by running something lower power then they will. If it gets the work done in time bound situations effectively they will use it, they don’t want to mess about with bios settings, they want out of the box to be just that… out of the box.
exactly, and a product needing a nerd to fine tune and test for stability and longevity is a failed product. ppl buy a product to work, not to tinker with, and buying the unlocked products to... tinker when they need to try squeeze out more power than stock, not lowering it... except some genius with a weird mindset...
Are we corporate clients here? Come on now. And I don't think corps would be buying this product anyways.
stuff and nonesense
Reputable
- Mar 10, 2020
- 508
- 425
- 5,270
If you buy on the bleeding edge, 9700x you expect to tinker, update bios etc… for now same with intel when their next new kit comes out. Corporates wait a while, test, validate and buy kit with a long term supply guarantee so that 18 months down the line they can get an identical model. They don’t want exceptions.
Next replacement cycle, rinse and repeat…
Fun fact is that even for Puget Systems, where you praised about their great tuning at their own lower power limits, says the 9700X and 9600X are efficient and despite Cinebench don't show performance gain, they are 5-10% faster than the 7000 series, "encourage caution when purchasing an Intel CPU right now until we see whether Intel’s promised microcode update addresses the issue." and "Overall, the Ryzen 7 9700X and Ryen 5 9600X processors are impressively efficient, but at their current price, are not the best in terms of value compared to AMD’s previous Ryzen 7000-series of processors. However, this is just for the lower-end Ryzen 5 and 7 processors, and we have yet to fully review the Ryzen 9 9950X and 9900X. The 9950X we are especially looking forward to, as it does not have a drop in TDP compared to the previous generation like the other models, which will give us a good idea of how Ryzen 9000 compares to the previous generation when running at the same power draw."
https://www.pugetsystems.com/labs/a..._9700X_and_Ryzen_5_9600X_for_Content_Creation
TRENDING THREADS
-
News Edward Snowden slams Nvidia's RTX 50-series 'F-tier value,' whistleblows on lackluster VRAM capacity- Started by Admin
- Replies: 84
-
Review Nvidia GeForce RTX 5080 Founders Edition review: Incremental gains over the previous generation
- Started by Admin
- Replies: 216
-
-
Review Nvidia GeForce RTX 5090 Founders Edition review: Blackwell commences its reign with a few stumbles
- Started by Admin
- Replies: 231
-
News Nvidia's RTX 50-series drivers feel half-baked, focus too much on MFG
- Started by Admin
- Replies: 76
-
Question Secondary SSD won't show up on windows after CPU upgrade
- Started by ic00042
- Replies: 19
Facebook
Twitter
Instagram