Putting the 27-inch i9 iMac thermal performance to the test

Posted:
in General Discussion edited August 2020
The recently refreshed 27-inch iMac offers a lot of power and capability for the price, in the same chassis that it has used for years. How does the new model's thermal performance under load stack up?

Credit: AppleInsider
Credit: AppleInsider


Compared to the aging iMac Pro, the 27-inch iMac with i9 processor delivers much more bang for a user's buck. But the 27-inch iMac has a different thermal design than the iMac Pro -- and it's one that is, ostensibly, less capable on paper.

To see the 27-inch iMac's thermal performance in action, we ran a model with a 3.6GHz 10th-generation Intel Core i9 processor through a variety of benchmarks. Here's what we found.

Benchmarks





We ran the 27-inch iMac through several types of benchmarks and speed tests to get a broad overview of the power of its 10th-generation Intel Core i9 chip. Our test device only has 8GB of RAM but is equipped with an AMD Radeon 5500XT with 8GB of video memory.

In a Blackmagic disk speed test, we saw write speeds of 1871.3 and read speeds of 2,365. Those are substantial gains over the entry-level 27-inch iMac, and is from parallelization with the increased drive size versus the lower-end model.

Firing up Geekbench and running a CPU benchmark, our 27-inch iMac clocked in with a 1256 single-core score and a 7958 multi-core score. That is quite a bit faster than the entry-level model, which scored 1141 and 5635 in the same test.

The 27-inch iMac with an AMD Radeon 5500XT with 8GB of RAM scored a 38318 in Geekbench compute testing. The improvements here over the entry-level model weren't as pronounced, but they're still solid.

In a Cinebench R20 render test, our iMac clocked in with a 5424 score. That's a huge improvement over the entry-level model, and mostly due to the 27-inch iMac's beefier processor.

What is "Thermal throttling"

All Intel processors demonstrate some level of so-called "Thermal throttling." This is by design, and not inherently negative.

Processors are created to run at a variety of different thermal design power (TDP) levels, depending on their intended market. Processors meant for desktop computers can afford to have a higher TDP, while notebooks, mobile devices, and tablets tend to use processors with lower TDP values, due to the difficulty of cooling the chip down in a physically constrained or power-limited environment.

While a processor is cool, it will exceed its normal operating clockspeed up to its boost speed. This generates more heat, of course. As that heat builds, the processor will slow down to help dissipate the heat in conjunction with the cooling system. It can even go lower than the normal operating clock speed if need be to prevent damage. That depression below normal operating clock speed is what is actually called "thermal throttling," and not the thermal regulation between the "turbo" speeds and the rated clock speed.

Our Cinebench R20 testing was performed in a typical office environment at 70F (21C). Running multiple Cinebench benchmarks in series with no cool-down period in between didn't cause the machine to cut the clock speeds below the normal operating speed.

The first Cinebench run didn't demonstrate any fan noise to speak of. After multiple runs, we saw sustained clock speeds around 4.3GHz through our testing, despite the temperature getting up to 100F (38C). The fans are clearly audible after two runs.

Before testing, we were originally concerned how well the 27-inch iMac would perform under load since it doesn't have the same thermal design as Apple's iMac Pro machine.

On the other hand, the 27-inch iMac i9 CPU at about 95W TDP is lower than the iMac Pro's Xeon processor at 115W TDP and up -- and this isn't even including the notoriously hot Vega 56 in the iMac Pro. The 27-inch iMac didn't adopt the iMac Pro cooling system, simply because it didn't have to.

Deals on Apple's new 27-inch iMac

Apple resellers are currently offering aggressive discounts on the new 27-inch iMac, with triple-digit savings already in effect on a variety of models.

For the latest prices and exclusive promo code savings, visit the AppleInsider 27-inch iMac Price Guide.
«1

Comments

  • Reply 1 of 24
    thttht Posts: 5,348member
    People call it thermal throttling, Intel calls it Turboboost. All 3 or 4 levels of turbo-boost they have these days.

    The chips are designed to operate reliably at temperatures lower than 100 °C. Since the vast majority of computers operate in 20 to 30 °C environments, Intel lets the processor run at higher clocks than the advertised base clocks until it reaches 100 °C, after which it will downclock to maintain lower temperatures than 100 °C for the given cooling systems and ambient environment. Ie, they are taking advantage of thermal capacitance. Thus the turbo clock speeds are advertised even though the processor spends very little time at that clock speed.

    Regarding the iMac versus iMac Pro, the tradeoff for nearly equivalent compute power between the two is still noise, as it was before. If low noise is important, paying the premium for the iMac Pro would be worth it to those folks.

    Not much longer to wait for the Apple Silicon versions though. I'd be amazed if Apple Silicon uses 100 W, let alone the 130 to 150 W that these 10-core CPUs in these new iMacs are using. I think Apple is targeting 50 to 60 W for the Apple Silicon SoCs in iMac equivalent machines.
    razorpitcat52joerock1234williamlondonwatto_cobra
  • Reply 2 of 24
    So... in the last 6 years we have gone from a single core score of 1067 to 1256, an improvement of less than 20% and multi-core scores have obviously only improved by doubling the number of cores to double the scores. This really shows how rubbish Intel have been, so much for Moore’s law lol


    cat52williamlondonwatto_cobra
  • Reply 3 of 24
    A welcomed "thermal" review.
    Mine arrives this coming Monday, replacing my trusty Late 2012 iMac 27 (fully loaded and maxed out when purchased then).

    Here's your new Mac.

    27-inch iMac with Retina 5K display

    Hardware:

    • 3.6GHz 10-core 10th-generation Intel Core i9 processor, Turbo Boost up to 5.0GHz
    • 8GB 2666MHz DDR4 memory
    • 4TB SSD storage
    • Radeon Pro 5700 XT with 16GB of GDDR6 memory
    • 10 Gigabit Ethernet
    • Nano-texture glass
    • Magic Mouse 2
    • Magic Keyboard - US English
    • Accessory Kit

    Ordered another 32GB's of memory, to drop into the two empty memory slots, as well
    jasonfont426williamlondonsunman42watto_cobra
  • Reply 4 of 24
    scartartscartart Posts: 201member
    tht said:
    People call it thermal throttling, Intel calls it Turboboost.
    They are different things. Thermal throttling is lowering the clock speed below base speed because it is too hot, caused by a manufacturer not following Intel’s guidelines on the amount of cooling required and it has a negative affect on performance. Turbo boost is running cores at a higher frequency when there is thermal capacity to do so, i.e. when not all cores are in use at the same time to boost performance. 

    aderutterwilliamlondonwatto_cobra
  • Reply 5 of 24
    StrangeDaysStrangeDays Posts: 12,768member
    scartart said:
    tht said:
    People call it thermal throttling, Intel calls it Turboboost.
    They are different things. Thermal throttling is lowering the clock speed below base speed because it is too hot, caused by a manufacturer not following Intel’s guidelines on the amount of cooling required and it has a negative affect on performance. Turbo boost is running cores at a higher frequency when there is thermal capacity to do so, i.e. when not all cores are in use at the same time to boost performance.
    It's not really based on failure to follow Intel guidelines. Intel has itself published info on when throttling is expected -- because it's part of the chip design when under a max load and the processor can't keep up. It happens; intensive gaming for example. In the real world thermal constraints cannot be avoided in our modern age of high-itensity computing. It's simply a matter of when, not if.
    williamlondonmacpluspluswatto_cobra
  • Reply 6 of 24
    dewmedewme Posts: 5,259member
    aderutter said:
    So... in the last 6 years we have gone from a single core score of 1067 to 1256, an improvement of less than 20% and multi-core scores have obviously only improved by doubling the number of cores to double the scores. This really shows how rubbish Intel have been, so much for Moore’s law lol



    You obviously do not understand the real problem or the underlying technology, i.e., computer architecture and physics. For general purpose CPUs the speedup potential for single threaded applications is directly related to the CPU frequency and memory access latency, which has not really improved significantly over the past couple of decades due to power and thermal dissipation issues and the reality of electrical engineering. There is no magic or rabbit that is going to be pulled from a hat that will suddenly improve the situation, it's a hard limit tied to the physics and underlying architecture that is limiting the speedup potential of every general purpose computer that has ever been built.

    Moore's Law is a heuristic, i.e., colloquial banter that makes for coffee shop and bar room conversation, but has no grounding in mathematical reality, where reality is defined by math, physics, and engineering. It is a cartoon based way of looking at the world as a spectator.

    Adding additional parallel computational units to a processor can indeed increase the speedup of a process, or reduce the time needed to get the answer to a problem, only if the algorithm, or solution to the problem, lends itself to distributing the computational workload needed to solve the problem across multiple computational engines. Unfortunately, the number of problems that lend themselves to this type of solution approach are not as prevalent as you might think. Think about a real world problem that you have encountered, like building a deck, and ask yourself whether have 8 or 10 additional people "helping" you solve the problem would have made your job easier or allowed you to complete your task quicker. Computers, and CPU cores, are no different than human brains, getting all of them working together in synchronicity to complete a single task isn't easy.

    Putting "Intel" and "rubbish" in the same sentence shows how little you truly understand about what it takes to deliver highly complex technology to market. Yes, it is fair to ding Intel on their laggardly performance in terms of rolling out new products using the most current fabrication technology. However, you totally fail to recognize the immensity of the challenges involved and the fact that the challenges involved are only within the purview of fewer than a handful of companies on the planet, There are very few among us who can legitimately criticize Intel for its performance when we consider the scope and scale of the problems they are solving. Weighing in as a passive and uninformed observer who is lacking technical insight, authority, and proven credentials to question their decisions carries absolutely no weight whatsoever.
    jasonfont426joerock1234williamlondonfastasleepwatto_cobra
  • Reply 7 of 24
    DuhSesameDuhSesame Posts: 1,278member
    4.3GHz isn't bad, although inferior than 10900K (which is 4.8GHz), it should have no problem beating most personal computers out there:

    https://store.steampowered.com/hwsurvey


    watto_cobra
  • Reply 8 of 24
    thttht Posts: 5,348member
    scartart said:
    tht said:
    People call it thermal throttling, Intel calls it Turboboost.
    They are different things. Thermal throttling is lowering the clock speed below base speed because it is too hot, caused by a manufacturer not following Intel’s guidelines on the amount of cooling required and it has a negative affect on performance. Turbo boost is running cores at a higher frequency when there is thermal capacity to do so, i.e. when not all cores are in use at the same time to boost performance.
    It's not really based on failure to follow Intel guidelines. Intel has itself published info on when throttling is expected -- because it's part of the chip design when under a max load and the processor can't keep up. It happens; intensive gaming for example. In the real world thermal constraints cannot be avoided in our modern age of high-itensity computing. It's simply a matter of when, not if.
    I agree with scartart’s definition of thermal throttling, and you two basically have the same definition.

    The big issue with Intel is they have been stuck on 14nm for 5 years. The only way they can really bring improved performance is to run the processors hotter. So, for the last two years of Intel processor generations, what system vendors have to do is design bigger cooling systems, pump more power into the CPU so that it can run at higher clocks. 

    For company like Apple who wants svelte, sleek, cool and quiet form factors, this trend is basically death to those desires. 

    There is going to be 1 more 14nm generation of Intel desktop processors with Rocket Lake. Ugh.
    watto_cobra
  • Reply 9 of 24
    thttht Posts: 5,348member
    dewme said:
    aderutter said:
    So... in the last 6 years we have gone from a single core score of 1067 to 1256, an improvement of less than 20% and multi-core scores have obviously only improved by doubling the number of cores to double the scores. This really shows how rubbish Intel have been, so much for Moore’s law lol



    You obviously do not understand the real problem or the underlying technology, i.e., computer architecture and physics. For general purpose CPUs the speedup potential for single threaded applications is directly related to the CPU frequency and memory access latency, which has not really improved significantly over the past couple of decades due to power and thermal dissipation issues and the reality of electrical engineering. There is no magic or rabbit that is going to be pulled from a hat that will suddenly improve the situation, it's a hard limit tied to the physics and underlying architecture that is limiting the speedup potential of every general purpose computer that has ever been built.
    ...
    Putting "Intel" and "rubbish" in the same sentence shows how little you truly understand about what it takes to deliver highly complex technology to market. Yes, it is fair to ding Intel on their laggardly performance in terms of rolling out new products using the most current fabrication technology. However, you totally fail to recognize the immensity of the challenges involved and the fact that the challenges involved are only within the purview of fewer than a handful of companies on the planet, There are very few among us who can legitimately criticize Intel for its performance when we consider the scope and scale of the problems they are solving. Weighing in as a passive and uninformed observer who is lacking technical insight, authority, and proven credentials to question their decisions carries absolutely no weight whatsoever.
    No, you can criticize Intel. They have fucked up, and if they were able to move to their 10nm on time and fabbing most of their processors with it, Apple likely wouldn't have switched to there custom processors. They changed CEOs because of the 10nm disaster. It now looks like they fucked up their 7nm fab, and they had to change their management again due to that. They now know they fucked up enough that they will go have TSMC fab some of their chips while they attempt to get their fabs in order.

    All of Intel's "Cove" CPU cores offer improved single threaded performance, but they need a fab with higher transistor density with good yield for Intel to make money with them. Those CPU cores would likely be hitting 1400 to 1600 GB5 points on single core on macOS. Nice 20% to 30% improvements, but those CPU designs have been waiting, waiting, waiting on 10nm to be productive. They won't be clocking higher, but have improved IPC. That is, they will outperform Skylake cores at lower clocks, but are designed for higher transistor density fabs.

    When the A14 comes out, sitting inside a phone, it's going to score somewhere between 1500 to 1600 GB5 points in single core. If the A14 implements SVE2, watch out, as that could get them to 1600 to 1800 GB5 single core scores. In a phone at about 5 Watts. In the meanwhile, the highest performance single core GB5 score on macOS for an Intel processor is somewhere around 1250 to 1300 at about 20 to 30 W.

    Really, you can criticize Intel and they rightly deserve it. TSMC not only took the lead, they are about to lap them.
    williamlondonmuthuk_vanalingamGG1watto_cobra
  • Reply 10 of 24
    This is a very interesting test. 

    Your mileage may vary depending on your usage. A test performed at Mac4ever, a French website, claims severe thermal throttling for the 10 core and this repeatedly. They advice to choose the 8 core which seems not to have the issue. Would it be possible for AppleInsider to mimic this test, just to ensure that the issue the French had was not due to a particular machine ?

    https://www.mac4ever.com/dossiers/156117_test-de-l-imac-27-2020-le-dernier-des-mohicans-video

    Their test results in Logic Pro X, Vue 2019 & Première Pro, result in the 8 core to be faster than the 10 core. 
    watto_cobra
  • Reply 11 of 24
    Mike WuertheleMike Wuerthele Posts: 6,813administrator
    OlivierHR said:
    This is a very interesting test. 

    Your mileage may vary depending on your usage. A test performed at Mac4ever, a French website, claims severe thermal throttling for the 10 core and this repeatedly. They advice to choose the 8 core which seems not to have the issue. Would it be possible for AppleInsider to mimic this test, just to ensure that the issue the French had was not due to a particular machine ?

    https://www.mac4ever.com/dossiers/156117_test-de-l-imac-27-2020-le-dernier-des-mohicans-video

    Their test results in Logic Pro X, Vue 2019 & Première Pro, result in the 8 core to be faster than the 10 core. 
    We saw that video. That has not been our experience.
    muthuk_vanalingamOlivierHRfastasleepwatto_cobra
  • Reply 12 of 24
    MacProMacPro Posts: 19,693member
    A welcomed "thermal" review.
    Mine arrives this coming Monday, replacing my trusty Late 2012 iMac 27 (fully loaded and maxed out when purchased then).

    Here's your new Mac.

    27-inch iMac with Retina 5K display

    Hardware:

    • 3.6GHz 10-core 10th-generation Intel Core i9 processor, Turbo Boost up to 5.0GHz
    • 8GB 2666MHz DDR4 memory
    • 4TB SSD storage
    • Radeon Pro 5700 XT with 16GB of GDDR6 memory
    • 10 Gigabit Ethernet
    • Nano-texture glass
    • Magic Mouse 2
    • Magic Keyboard - US English
    • Accessory Kit

    Ordered another 32GB's of memory, to drop into the two empty memory slots, as well
    I could be wrong or misunderstanding you, but, I don't think you will have two empty slots.  You will have to remove the 4 x Whatevers and replace them with 4 x 16s.  I bought the lowest RAM and upgraded myself.  I went to 64 GB by buying 3rd party 4 x 16s and have the Apple stock 4 x 2s in a drawer.
    edited August 2020
  • Reply 13 of 24
    sunman42sunman42 Posts: 254member
    MacPro said:
    [snip]
    Ordered another 32GB's of memory, to drop into the two empty memory slots, as well
    I could be wrong or misunderstanding you, but, I don't think you will have two empty slots.  You will have to remove the 4 x Whatevers and replace them with 4 x 16s.  I bought the lowest RAM and upgraded myself.  I went to 64 GB by buying 3rd party 4 x 16s and have the Apple stock 4 x 2s in a drawer.

    I believe BuffyzDead is correct: the base memory configuration is 2 x 4 Gbyte of memory, leaving two slots empty, which could be filled with 2 x 16 Gbyte, for a total of 40 Gbyte.

    watto_cobra
  • Reply 14 of 24
    sunman42sunman42 Posts: 254member
    A welcomed "thermal" review.
    Mine arrives this coming Monday, replacing my trusty Late 2012 iMac 27 (fully loaded and maxed out when purchased then).

    Here's your new Mac.

    27-inch iMac with Retina 5K display

    Hardware:

    • 3.6GHz 10-core 10th-generation Intel Core i9 processor, Turbo Boost up to 5.0GHz
    • 8GB 2666MHz DDR4 memory
    • 4TB SSD storage
    • Radeon Pro 5700 XT with 16GB of GDDR6 memory
    • 10 Gigabit Ethernet
    • Nano-texture glass
    • Magic Mouse 2
    • Magic Keyboard - US English
    • Accessory Kit

    Ordered another 32GB's of memory, to drop into the two empty memory slots, as well
        Since this is exactly the config (well, minus the $500 Nano-texture glass) I'm looking at ordering when AppleCare on my 2017 iMac runs out in October, I'd be really interested in seeing Cinebench stress (fan, temperature, power consumption, and CPU clock) results. Hope you're planning on running those tests, and sharing!
    watto_cobra
  • Reply 15 of 24
    MacProMacPro Posts: 19,693member
    Completely unscientific but may be of interest Mac gamers.  

    I have set up a side by side;  

    Dell XPS 8930 PC which has 64 GB RAM, 8th Gen Intel Core i7-8700, with Nvidia GTX 1080 into a 4K Dell 27" monitor with Windows 10 Pro  (a mid-level gaming desktop IMHO).

    iMac 27" i9, 5K, 64 GB RAM with the Radeon Pro Vega 48 8 GB, booting into Windows 10 Pro.  

    Both can run Red Dead Redemption II and GTAV in 4K with quality maxed out at around 60 f.p.s. and the fans seem pretty much the same noise level... when they come on, which is intermittent on both.  Neither get overly hot.  Frankly, I can't tell any difference between them in use.   The iMac can go up to 5K without any frame rate difference, the Dell is maxed out at Ultra-HD but visually both look amazing.

    I am currently installing Microsoft Flight Simulator 2020 on the iMac 5K, hoping that with run as well as the Dell.
    muthuk_vanalingamwatto_cobra
  • Reply 16 of 24
    Good to know it doesn't have thermal throttling issues, although other reviewers have reported the opposite. I would have liked to hear more about how loud the fans are. How distracting is the noise? How much throttling do you get if you cap the fan speed to a quiet level?
    edited August 2020 watto_cobra
  • Reply 17 of 24
    dewmedewme Posts: 5,259member
    tht said:
    dewme said:
    aderutter said:
    So... in the last 6 years we have gone from a single core score of 1067 to 1256, an improvement of less than 20% and multi-core scores have obviously only improved by doubling the number of cores to double the scores. This really shows how rubbish Intel have been, so much for Moore’s law lol



    You obviously do not understand the real problem or the underlying technology, i.e., computer architecture and physics. For general purpose CPUs the speedup potential for single threaded applications is directly related to the CPU frequency and memory access latency, which has not really improved significantly over the past couple of decades due to power and thermal dissipation issues and the reality of electrical engineering. There is no magic or rabbit that is going to be pulled from a hat that will suddenly improve the situation, it's a hard limit tied to the physics and underlying architecture that is limiting the speedup potential of every general purpose computer that has ever been built.
    ...
    Putting "Intel" and "rubbish" in the same sentence shows how little you truly understand about what it takes to deliver highly complex technology to market. Yes, it is fair to ding Intel on their laggardly performance in terms of rolling out new products using the most current fabrication technology. However, you totally fail to recognize the immensity of the challenges involved and the fact that the challenges involved are only within the purview of fewer than a handful of companies on the planet, There are very few among us who can legitimately criticize Intel for its performance when we consider the scope and scale of the problems they are solving. Weighing in as a passive and uninformed observer who is lacking technical insight, authority, and proven credentials to question their decisions carries absolutely no weight whatsoever.
    No, you can criticize Intel. They have fucked up, and if they were able to move to their 10nm on time and fabbing most of their processors with it, Apple likely wouldn't have switched to there custom processors. They changed CEOs because of the 10nm disaster. It now looks like they fucked up their 7nm fab, and they had to change their management again due to that. They now know they fucked up enough that they will go have TSMC fab some of their chips while they attempt to get their fabs in order.

    All of Intel's "Cove" CPU cores offer improved single threaded performance, but they need a fab with higher transistor density with good yield for Intel to make money with them. Those CPU cores would likely be hitting 1400 to 1600 GB5 points on single core on macOS. Nice 20% to 30% improvements, but those CPU designs have been waiting, waiting, waiting on 10nm to be productive. They won't be clocking higher, but have improved IPC. That is, they will outperform Skylake cores at lower clocks, but are designed for higher transistor density fabs.

    When the A14 comes out, sitting inside a phone, it's going to score somewhere between 1500 to 1600 GB5 points in single core. If the A14 implements SVE2, watch out, as that could get them to 1600 to 1800 GB5 single core scores. In a phone at about 5 Watts. In the meanwhile, the highest performance single core GB5 score on macOS for an Intel processor is somewhere around 1250 to 1300 at about 20 to 30 W.

    Really, you can criticize Intel and they rightly deserve it. TSMC not only took the lead, they are about to lap them.
    Per my original comments we are in total agreement that Intel's laggardly execution on the fabrication side is hurting their reputation and crushing their bottom line. The point I am making is in regard to any "Moore's Law" type of expectations around single threaded performance, which are whacky. Your point about the potential 20% raw improvements with Intel's Cove cores reinforces my point exactly. To classify Intel as "rubbish" is unwarranted, especially in terms of single threaded performance in high-end desktop platforms.

    Intel obviously got complacent and rode their cash cow way way too long and also got burned by some negative security artifacts lingering in legacy designs. But at their core engineering and design level, Intel still has the chops to be very competitive and stay in the fight against TSMC and will likely dominate in certain problem domains. Intel using TSMC's fab capability is a pragmatic stop-gap measure and is really no different than Apple leaning heavily on Samsung for some of their core components across Apple's product lines. Doing so clears some of the workload from Intel's plate to free up much needed bandwidth. It's not Intel waving a white flag by any stretch of the imagination. 
    watto_cobra
  • Reply 18 of 24
    flydogflydog Posts: 1,117member
    aderutter said:
    So... in the last 6 years we have gone from a single core score of 1067 to 1256, an improvement of less than 20% and multi-core scores have obviously only improved by doubling the number of cores to double the scores. This really shows how rubbish Intel have been, so much for Moore’s law lol


    About all you've accomplished with your post is show the world how ignorant your are.  

    First, Moore's Law says nothing about performance or benchmark scores.   I states that the number of transistors doubles every two years.  The number of transistors alone does not dictate performance.  

    Second, single core benchmarks mean very little because most modern computers rely on multitasking across multiple cores to lift the heavy loads.  Adding cores is more power efficient than simply increasing clock speed.

    Third, the number of cores and clock speed also mean very little. The 2008 Mac Pro and 2013 iMac Pro were both available with a Xeon 8-core 3.2 Ghz processors. The 2017 is 3x faster in multicore, while using less power.  So contrary to your ridiculous uninformed post, we've come a long way.
    fastasleepwatto_cobra
  • Reply 19 of 24
    thttht Posts: 5,348member
    dewme said:
    The point I am making is in regard to any "Moore's Law" type of expectations around single threaded performance, which are whacky. Your point about the potential 20% raw improvements with Intel's Cove cores reinforces my point exactly. To classify Intel as "rubbish" is unwarranted, especially in terms of single threaded performance in high-end desktop platforms. 
    Your expectations for single thread performance are too pessimistic. Perhaps aderutter should not have mentioned anything about Moore's Law, but the 20% single core performance gain across 5 to 6 years is primarily the result of Intel not being about to fab desktop processors on their 10 nm fab. If they were able to, single core performance improvements from 2015 to 2020 would be in the 50% to 70% range.

    Tiger Lake GB5 scores are hovering around 1400 to 1600 single core points. They could have had that last year if 10nm was working, and they wouldn't need to backport Willow Cove to 14nm for Rocket Lake to maintain a semblance of parity with the mobile chips. Willow Cove on 10nm at desktop TDPs could hit 1600 to 1800 GB5 single core points. 

    I would describe Intel's last two generations of desktop processors as rubbish, figurative turds. They have turbo power consumption levels of 150 to 250 W and added 125 W SKU line! All that for not much performance gain. That's nothing for Intel to be proud of, and I bet Intel OEMs aren't all that happy about it either.
    watto_cobra
  • Reply 20 of 24
    DuhSesameDuhSesame Posts: 1,278member
    tht said:
    dewme said:
    The point I am making is in regard to any "Moore's Law" type of expectations around single threaded performance, which are whacky. Your point about the potential 20% raw improvements with Intel's Cove cores reinforces my point exactly. To classify Intel as "rubbish" is unwarranted, especially in terms of single threaded performance in high-end desktop platforms. 
    Your expectations for single thread performance are too pessimistic. Perhaps aderutter should not have mentioned anything about Moore's Law, but the 20% single core performance gain across 5 to 6 years is primarily the result of Intel not being about to fab desktop processors on their 10 nm fab. If they were able to, single core performance improvements from 2015 to 2020 would be in the 50% to 70% range.

    Tiger Lake GB5 scores are hovering around 1400 to 1600 single core points. They could have had that last year if 10nm was working, and they wouldn't need to backport Willow Cove to 14nm for Rocket Lake to maintain a semblance of parity with the mobile chips. Willow Cove on 10nm at desktop TDPs could hit 1600 to 1800 GB5 single core points. 

    I would describe Intel's last two generations of desktop processors as rubbish, figurative turds. They have turbo power consumption levels of 150 to 250 W and added 125 W SKU line! All that for not much performance gain. That's nothing for Intel to be proud of, and I bet Intel OEMs aren't all that happy about it either.
    Skylakes?  Intel bumps the frequency thanks to the steady improvements of the 14nm.  If the frequency remains the same, however...
Sign In or Register to comment.