Intel Xeon 28 Cores / 56 Threads

[yathis]

Businesses will buy it

[Drak3]

1 hour ago, RadiatingLight said:

all energy that goes into a CPU gets outputted as heat. basic thermodynamics

False. All energy is converted and not destroyed, but not all energy sent to a CPU becomes heat, electrical signal is sent back to RAM, PCIe, and all other I/O in system. Those signals will eventually be translated by devices to be usable to users, these include: graphics processors, sound processors, monitors, and speakers. Otherwise, a computer could not compute, it'd be a space heater.

For Intel processors, it's commonly estimated that TDP is 60-70% of total power draw under the load used to specify that TDP.

 

1 hour ago, RadiatingLight said:

Intel just puts all their X99 CPUs under a broad roof, 140W TDP, even though the TDP of a 6900K is probably more like 100W or less.

Again, false. Intel's TDP measurements are done under specific workloads. Heavier loads than consumers typically use. I can drive any and every X99 CPU to its TDP, and then some, with minimal effort, I just need to stack Prime95 with another heavy benchmark. But that's synthetic load that is dissimilar to the actual load my system sees. I could also do that with my i5-4460, GTX 1080, or RX470. Hell, I can do it with RAM, the hardest part would be tracking down the TDP of my consumer RAM.

 

TDP is not the highest heat output that components put out, it's the heat output at a high load, as determined by the telemetry data that every processor manufacturer collects. With some effort, they can be surpassed.

[Droidbot]

4 hours ago, RS2007GOD said:

Yea I was expectibg around that but 15k is another level. 

that's what you pay for cutting edge

 

[Bananasplit_00]

2 hours ago, RadiatingLight said:

Um.. not that's not how it works.

all energy that goes into a CPU gets outputted as heat. basic thermodynamics

Intel just puts all their X99 CPUs under a broad roof, 140W TDP, even though the TDP of a 6900K is probably more like 100W or less.

i really doubt the 6900K is 100W, its a 8 core chip. probably more like 120W also im pretty sure not all energy is put out as heat, i havent read thermodynamics yet but that sounds a bit strange to me, wouldent the heat just be efficiency losses? i find it strange that every single watt going in becomes heat.

[Bananasplit_00]

47 minutes ago, Droidbot said:

that's what you pay for cutting edge

 

yay for early adopter tax!  

[RadiatingLight]

7 hours ago, Bananasplit_00 said:

i really doubt the 6900K is 100W, its a 8 core chip. probably more like 120W also im pretty sure not all energy is put out as heat, i havent read thermodynamics yet but that sounds a bit strange to me, wouldent the heat just be efficiency losses? i find it strange that every single watt going in becomes heat.

Well one of the main laws of thermodynamics is that energy cannot be created or destroyed.

all energy that goes into a CPU in the form of electricity, must be transformed into another form of energy. it's possible that a very small percentage of the energy goes out in the form of a very weak magnetic field, or something like that, but in principle, all energy that goes in = heat that goes out.

 

efficiency is just how much work a processor can do with a certain amount of electricity.

 

like old CPUs can do a small amount of work with 10W of power, while modern CPUs can do a lot of work with 10W of power. still the same heat output tho.

[Coaxialgamer]

13 hours ago, RadiatingLight said:

What?

 

I knew TDP was calculated differently, but there's no fucking way it uses 160W. you'd essentially require either an AIO or a NH-D15 (Or other comparable air cooler) to cool a 160W TDP CPU.

Well you can't use tdp as an accurate measure of anything tbh. The same value is often used for entire families of products, and their isn't a universally agreed way to calculate it. 

But because 99.9% of the energy a cpu uses is dispersed as heat, it's safe to say that power consumption = actual heat output. 

[Coaxialgamer]

8 hours ago, Drak3 said:

False. All energy is converted and not destroyed, but not all energy sent to a CPU becomes heat, electrical signal is sent back to RAM, PCIe, and all other I/O in system. Those signals will eventually be translated by devices to be usable to users, these include: graphics processors, sound processors, monitors, and speakers. Otherwise, a computer could not compute, it'd be a space heater.

 

8 hours ago, Drak3 said:

 

The energy in the form of I/O is minimal at best. The majority is still lost as heat. 

Even the signals sent to say RAM, cache etc are all lost as heat as soon as system power is shut off. 

 

 

[TheRandomness]

-Thread moved to CPUs, Motherboards, and Memory. @Energycore Don't you think it'd suit this subforum better?

[yathis]

16 hours ago, RadiatingLight said:

all energy that goes into a CPU gets outputted as heat

not all

 

[porina]

14 hours ago, Drak3 said:

Intel's TDP measurements are done under specific workloads. Heavier loads than consumers typically use. I can drive any and every X99 CPU to its TDP, and then some, with minimal effort, I just need to stack Prime95 with another heavy benchmark. But that's synthetic load that is dissimilar to the actual load my system sees. I could also do that with my i5-4460, GTX 1080, or RX470. Hell, I can do it with RAM, the hardest part would be tracking down the TDP of my consumer RAM.

Further to this point, Prime95 and similar software use FMA3 (AVX2) instructions heavily, and that's where the heat really turns up. When AMD designed Ryzen, they specifically chose not to implement the same level of hardware resource in these instructions saying that most people don't use it. In this area, it is about half the IPC compared to recent Intel. Thus in a similar core/clock configuration, they will do less work, and use less power. They can use it to selectively tell half-truths making it sound like their processors are much more efficient, whereas in most cases if you compared performance per watt, it is pretty close. This was even visible in the press pre-release demos they gave.

[RadiatingLight]

12 minutes ago, yathis said:

not all

 

so where does the other energy go lol

[yathis]

its used by the cpu

 

1 minute ago, RadiatingLight said:

so where does the other energy go lol

 

its like an electric motor, 100% of the energy is not wasted in heat....... its actually used.

The heat is inefficiency. Everything has loss to it, just a matter of how much.

 

[RadiatingLight]

1 minute ago, yathis said:

its used by the cpu

 

 

its like an electric motor, 100% of the energy is not wasted in heat....... its actually used.

The heat is inefficiency. Everything has loss to it, just a matter of how much.

 

You obviously don't understand thermodynamics

 

electric motors output some of their energy in heat, and some of their energy as motion. that motion is usually used to move something forward, which still retains the kinetic energy. taking a car as an example, when it stops, the kinetic energy is again converted to heat, so eventually, all the energy that goes into a motor ends up as heat.

 

just look it up. or provide a source for your claims.

[yathis]

3 minutes ago, RadiatingLight said:

energy in heat, and some of their energy as motion.

yes

 

and

 

same goes for cpu, but instead of motion, its "work"

 

Your statement of claim was that 100% of the energy input is wasted output heat.

Which is obviously not true at all. Sure there is heat, that is why they got coolers and fans. Which you know.

 

Source to my claim - Common Sense

 

 

LMFAO

[DrMikeNZ]

13 hours ago, Bananasplit_00 said:

i really doubt the 6900K is 100W, its a 8 core chip. probably more like 120W also im pretty sure not all energy is put out as heat, i havent read thermodynamics yet but that sounds a bit strange to me, wouldent the heat just be efficiency losses? i find it strange that every single watt going in becomes heat.

It depends on the clock speed and workload.

At stock speeds with handbrake encoding workload, my 6900K drew around 60W. Overclocked to 4.0GHz it drew around 140W encoding, and with Prime95 workload my 6900K draws around 200W.

My R7 1700 only draws around 65W at 3.8GHz with handbrake encoding and Prime 95 workloads, although it does take about twice as long to complete the tasks.

 

2 minutes ago, RadiatingLight said:

so where does the other energy go lol

A tiny portion could be lost as non-thermal electromagnetic radiation.

[Energycore]

40 minutes ago, TheRandomness said:

-Thread moved to CPUs, Motherboards, and Memory. @Energycore Don't you think it'd suit this subforum better?

:PPPPP

[SpaceGhostC2C]

21 hours ago, Hunter259 said:

https://www.pcper.com/reviews/Processors/AMD-Ryzen-7-1800X-Review-Now-and-Zen/Power-Consumption-and-Conclusions

read that slightly wrong. It is more like 130-140 watts under load. 

They go by kill-a-watt measures. Meaning, they simply don't know what the CPUs' power consumption is. 

 

2 hours ago, yathis said:

 

same goes for cpu, but instead of motion, its "work"

Engine: Potential energy transforms into kinetic energy, kinetic energy transforms into heat (hence quite brakes heat up). 

 

CPU: potential energy transforms into...? Hint: you have to fill the dots with an actual form of energy. "Work" isn't one. 

 

2 hours ago, yathis said:

Source to my claim - Common Sense

Maybe it's time to try with science instead. 

[Drak3]

3 hours ago, porina said:

When AMD designed Ryzen, they specifically chose not to implement the same level of hardware resource in these instructions saying that most people don't use it.

To be completely fair, AVX isn't used much in the consumer world yet, and as far as I can see, it's mostly used for stress testing.

[porina]

8 hours ago, Drak3 said:

To be completely fair, AVX isn't used much in the consumer world yet, and as far as I can see, it's mostly used for stress testing.

My concern is by taking this path it allows AMD to claim a lower TDP while being performance competitive in other areas. This may lead Intel to cripple their future products in a race to the bottom. OTOH I'm still impatiently waiting for AVX512 to go into consumer CPUs...

[DrMikeNZ]

9 hours ago, SpaceGhostC2C said:

CPU: potential energy transforms into...? Hint: you have to fill the dots with an actual form of energy. "Work" isn't one. 

Work is an actual form of energy. Just multiply the force applied by the distance traveled. Won't amount to much of anything in a CPU though.

8 hours ago, Drak3 said:

To be completely fair, AVX isn't used much in the consumer world yet, and as far as I can see, it's mostly used for stress testing.

Prime95 is a legitimate program for recreational mathematics used by quite a lot of people, not (just) for stress testing.

AVX is very useful for science and other fields of maths as well.

3 minutes ago, porina said:

My concern is by taking this path it allows AMD to claim a lower TDP while being performance competitive in other areas. This may lead Intel to cripple their future products in a race to the bottom. OTOH I'm still impatiently waiting for AVX512 to go into consumer CPUs...

With how terrible Ryzen has been for me in my workload, I will not be getting any more AMD hardware for a while. If Intel cripples its consumer hardware then I will just have to go for their enterprise grade processors.

The number of consumers that would benefit from AVX512 is pretty small, and it would take a miracle for that to happen, but we can hope.

[porina]

1 hour ago, DrMikeNZ said:

With how terrible Ryzen has been for me in my workload, I will not be getting any more AMD hardware for a while. If Intel cripples its consumer hardware then I will just have to go for their enterprise grade processors.

That's my concern also. It may be in the future that for any useful performance features you might need to go to Xeon CPUs as standard, as consumer level kit is optimised otherwise. The problem I have with it is obviously Xeons will usually come at a cost premium. For now, the right i5 or i7 provides good value over similar performance Xeons, although there has been some overlap in the past.

1 hour ago, DrMikeNZ said:

The number of consumers that would benefit from AVX512 is pretty small, and it would take a miracle for that to happen, but we can hope.

The number of potential users of AVX512 will be similar to the numbers of those using AVX2 already. Software will need to be updated to make use of it, but if software can use AVX2 performance now, they're unlikely to turn away the further gain from AVX512. My fear is that if AVX in general becomes a "high end only" thing, it will get less traction than it already does.

[yathis]

CPU's dont do work they are lazy you are right.

Bunch of microscopic moving switches dont do nothing, stand still, nothingness.