In autumn without realizing the sun and rain, longing spreads like mountains~~
Hello everyone I am Yuna, a hardware and digital enthusiast. Today, let us regain our memory and study an old CPU - Core i7-2600K.
Core i7-2600K was born on January 9, 2011, and it has been more than 10 years ago. As a classic "Shen U", it not only has strong performance and considerable overclocking potential, but its classic Sandy Bridge architecture has the influence to this day, and it can be called a "http://www.. The legend of living ".
When it comes to 2022, how much combat power is left? How glorious would it be if you owned it back then?
Now let’s experience it up close and first uncover the answer to the first question.
History
Although Intel launched the epic Core 2 in 2006, which surpassed AMD in terms of advanced architecture, it has not completed the crushing trend for several years. The reason is that Core 2 has the constraint of front-end bus bandwidth, which leads to the effect of dual-channel 667~800MHz DDR2, no matter how good the memory is, it can only run into the dual-channel 667~800MHz DDR2. Nehalem, which was launched at the end of 2008, was also the so-called "first generation Core ", solved this problem by integrating IMC, but in the end it was still a huge 45nm process; Westmere successfully ticked to the 32nm process, but the upper limit was only 2 cores and 4 threads "i3 level". During this period, AMD relies on the two major magic weapons of X3/core opening, Tianji horse racing, and is still quite popular in the cost-effective market.
finally swung a fatal sword and achieved the glory of "i3 silent seconds" was the second generation Core Sandy Bridge after the tock. Although there are reasons why AMD's own new architecture fails, Sandy Bridge's success is far more than that. It has a milestone -
- The revolutionary Ringbus 2Ringbus: each core has an independent L1+L2 cache; each core corresponds to a cache area, and can access all L3 caches; all cores maintain a consistent short communication distance between the L3 cache; integrated core graphics, media engines, system assistants, etc. are also connected to Ringbus. Ringbus laid the foundation for the design ideas of many generations of CPUs since then, including rival AMD. was still using Ringbus until the 13th generation Core released a few days ago.
- High efficiency AVX instruction set. It enhances the performance of many CPU including integer pipelines, floating point throughput, and FMA high-precision computing efficiency, and provides benefits for the development of fields other than CPU. Today, many game/video editing software will use AVX acceleration.
- promotes the popularization of USB 3.0. Strictly speaking, this credit should belong to AMD's A75 chipset, but considering the huge installed capacity of the second generation Core, it is not an exaggeration to say that it is driven by it. USB 3.0 has been transformed from the previous generation 2.0, driving significant improvements in peripherals/storage, benefiting PC users to this day.
Sandy Bridge has such a high starting point, and AMD is also very "cooperating", which in the end has put Intel in the reputation of "toothpaste factory" - everyone can see this from the subsequent review content. Today's protagonist, i7-2600K, is the star of the Sandy Bridge family.
(on the left is its younger brother Pentium G2030, which was also a popular U back then)
CPU-Z
i7-2600K belongs to the LGA1155 platform, and the third generation Core Ivy Bridge also uses this platform. In the future, it will be "technology is based on platform exchange".
I chose Maximus V Gene (referred to as M5G, Z77 chipset) from ROG "pure blood" as its car. I suddenly realized that this was the first ROG motherboard in my life.
i7-2600K inherits the specifications of the predecessor i7-870: core + 8 threads + 8M Level 3 cache . During the same period, the i5 family was 4+4+6, and the i3 family was 2+4+3 or 4. This set of settings continued to the 7th generation Core in 2016. Its basic frequency is 3.4GHz, single-core/full-core acceleration frequencies are 3.8/3.6GHz respectively, and it does not lock the frequency double. It has powerful overclocking capabilities. In fact, looking back today, it is also the last generation of CPUs that can make you feel very comfortable. The overclocking basic disk of
generation Core is generally 4.4GHz; 4.5~4.6 can be achieved through pressurization when the motherboard/heat dissipation is good, and above 4.7 tests the CPU physical fitness. Although the third generation Core has ticked to 22nm again, its high-frequency rushing capability has not been significantly improved - the most serious thing is that its silicon grease replaces the thermal conductivity design of brazed , making it worse than the second generation Core, and the result is generally enough to surpass the basic plate.
My i7-2600K is of medium upper body shape, and can overclock to 4.8GHz of the full core at 1.416V and pass through the roasted chicken. At this time, the CPU packaging power consumption is above 130W, under room temperature of 27℃ + Yajun G3 suppression, the temperature of single baked FPU is 82~84℃, and some subsequent tests are carried out at this frequency.
i7-2600K also integrates Intel's milestone core display HD 3000. Previously, the first generation of Core's HD Graphics was only two separate chips that packaged with the CPU, and it cannot be considered a real "core display". The HD 3000 has 12 sets of execution units (EUs), and its performance is equivalent to the entry-level GT 610 in the same year. It can output up to 2K 60Hz images with the DP interface - so don't say that "xx graphics card has a display that doesn't move", it's not just driving when playing games.
Due to the weak performance of HD 3000, this article will no longer add any effect to it. subsequent tests will block it . The driver interface is full of age~~
compared to the later generations of Arc's independent display -
test platform
The opponent I found for i7-2600K is a modern architecture CPU - Ryzen 7 5700G. It was released in 2021 and is 10 years younger than i7-2600K! The process has improved to TSMC's 7nm, and the scale has doubled to 8 cores and 16 threads.
In order to make the comparison test more "interesting", I added some whole-work operations and finally divided the contestants into 4 groups -
- 600K silent frequency, the memory is the default 1333MHz 9-9-9-24 - This is 2600K in the original "no toss" state;
- 600K overclocked to 4.8GHz for the full core, and the memory is overclocked to 2133MHz 9-10-10-24—represents the top 1155 platform;
- 700G shuts down half of the core to make it 4 cores and 8 threads with 2600K, and locks the frequency to 4.4GHz—so that it can basically represent i7-7700K overclocked to 5.0GHz of all cores ( is referred to as fake 7700K overclocking ), a junior of 2600K;
- 700G default setting, open all core (because the original cooling cannot be dissipated, I will no longer open PBO) - represents the current mainstream CPU .
comparison group memory, both of which are recommended in daily life, , which I write configuration list, and recommends more frequently. It is overclocked to 3600MHz 16-18-18-36.
One more mouth i7-2600K memory partner Samsung gold bar , this memory is also the peak of DDR3, and it can directly pull 2133 (but the bulldozer platform seems to be only 1933). Samsung is relying on it to defeat Erbida and then sit on the throne of the memory grain leader. After that, Samsung gold bars slowly disappeared from the DIY market, and only existed in laptops and server products, and were filled with a large number of fake bars. Generally speaking, it is not recommended that students who want to pick up DDR3 garbage to buy Samsung.
There is a saying that the authentic Samsung gold bar PCB board has a little yellow color below, and the green one on it is a fake bar.I collected both of these notes and were mixed. I found that they could run 2133CL9 steadily - so there is no way to verify it. It is best to add if you have any knowledge.
groups of contestants use RX 590 Ultra Platinum Aurora Special Edition independent graphics card released at the end of 2018. It is based on GF 12nm process, 36CU has a total of 2304 stream processor , and its performance is extremely close to the previous generation of gaming console King X1X. In the era of i7-2600K, even the famous "nuclear bomb" GTX 690 cannot surpass it with the ideal dual-core optimization. As a partner of 1080P, it should not be dragged down.
running scores
convention summary is placed in front, and detailed description is placed in the back. Students who are too long can only read the summary and then fast forward.
First, there is an extra wave of popular Entertainment Master Classic Version , because the absolute value of the score is too high, it was not included in the chart.
Default frequency + default memory group, the CPU score of i7-2600K is about 7W, which is very similar to the first generation of God U E3-1230V2.
CPU+memory dual over-group, overclocked to the full core 4.8GHz 2600K score exceeded 9W. This result even reached the new generation of gaming god U i5-8400/9400F, which has no longer "squeezing toothpaste" after many years.
, but even with the bonus of high-frequency memory + extremely low latency, the score of RX 590 graphics card is not much improved. It seems that in the early 20W and the two comparison groups - shows that even the i7-2600K + DDR3 memory + PCIe 2.0 with the silent frequency is enough to bring RX 590; in the upper limit of GTX 690, it was even more killing .
One more note: Ryzen 5000 series APU, the desktop has a complete PCIe 3.0*16, which is different from the mobile terminal.
space is limited. If there is no special explanation below, only the screenshot of CPU+memory double supergroup is placed.
~3rd generation Core's CPU-Z running score is very interesting, basically it is divided by 10. The single-core/multi-core scores of the i7-2600K silent frequency are very similar to the previous generation of God U E3-1230V2; after overclocking to 4.8GHz of all cores, it can be comparable to the i7-6700K/7700 years later! It perfectly explains what is ()... Compared with the fake 7700K overclocking, the single-core performance is 27% behind - the IPC of Skylake architecture is about 25% higher than Sandy Bridge. Considering the high-frequency attenuation factor, it is indeed very consistent with the theoretical gap. However, the multi-core with real 7700K overclocking should be only around 3000, and the hyperthreading efficiency of Ryzen is still a little higher.
Cinebench R15 does not use the AVX module, it is a relatively friendly project for old U. The i7-2600K scores a single core 135 multi-core 670. After overclocking, it can be increased to a single core 170 multi-core 850. The improvement of 5%+ level is equivalent to an additional core out of thin air - the so-called "the U that can make you feel super comfortable". The subsequent CPU overclocking is generally more of a sense of ritual.
R15 is more inclined to AMD. After overclocking, the fake 7700K runs to single-core 235, which really should be 215~216. After looking back, it is about 27%.
Cinebench R23 Because it can use the AVX2 module, the i7-2600K, which only AVX does not have AVX2, is at a disadvantage. The silent order core was run for nearly 20 minutes before it finished; after overclocking, its single-core score reached 920, and the gap with the fake 7700K widened to more than 50%, which is obviously due to AVX2.
Even so, the multi-core score of i7-2600K 600+ after overclocking is still considerable, and it has reached the level of the popular standard gaming laptop i5-10300H in 2020. Considering that many manufacturers still used 10300H with 1660Ti or even 2060 graphics cards at the beginning, my matching today is completely unacceptable to "low U-high graphics".
DMark11 is an old-fashioned test project. Because it happens to be the same as i7-2600K, I moved in to entertain the entertainment industry. The previous R15 has Master Lu . The physical score after overclocking of
i7-2600K exceeded 10,000, and the increase in the frequency of the silent frequency reached nearly 40%! However, the gap between the two comparison groups is quite large.
Now the mainstream DMark TimeSpy, its physical score will use the AVX2 instruction set, so i7-2600K is far behind the comparison group.
But the only reason is that under the premise that single-core performance is at a disadvantage, graphic scores have both exceeded the comparison group ! Although the gap is small and close to the error level, I ran several times and the protagonist group performed better. Only by putting this RX 590 on my R9-5900X main machine can we overtake it - perhaps we can simply and roughly understand it as: Intel's CPU is indeed easier to drive the graphics card to play than AMD.
FF15 benchmark test is similar to Time Spy, 080P+standard image quality , i7-2600K silent frequency defeated the 4 core and 8 threads comparison group, and after overclocking, it also defeated the complete body comparison group. In fact, using R9 5900X to bring RX 590 can only run to 6600 points.
Compared with the many games below, FF15's test script is actually closer to the actual game . It covers many real game content such as driving, riding, fishing, fighting, summoning, camping, etc., rather than just a simple cutscene demonstration. This result is a powerful indication that when the CPU does not have a full load bottleneck, paired with a reasonable graphics card, the gaming experience of the i7-2600K is still very close to contemporary U.
Game
I selected 14 games, and the first 10 happened to be released from 2011 to 2020, one per year, and each optimization is generally at a normal level in the corresponding year, and they also have benchmarks to eliminate interference caused by improper operation. The resolutions of are the most general 1920*1080, which is just the same as the development of gaming hardware requirements in the past 10 years.
minimum frame number is 1% low. After all, 0.1% low is too difficult to perceive by ordinary players, and the variables are also large. Some screenshots show 0.1%low, and 1%low is my extra count.
Only frame count screenshots will be taken. Next article will add the game introduction and screenshots.
011 is the year when i7-2600K was born. The game I chose is Dust 3, with the highest image quality, 4x anti-aliasing is enabled.
i7-2600K is not satisfied with the graphics card under the silent frequency, but the average frame rate of 164 and the lowest frame rate of 119 is still very smooth. High-flash monitors were not very popular back then. After overclocking, the graphics card can be roughly full like the two comparison groups, and the minimum frame rate is approximately equal to the average frame rate before overclocking.
012 selected Hot-blooded rogue , Hong Kong version of GTA. The highest image quality of this game unexpectedly took the graphics card, so it was reduced to high image quality.
i7-2600K silent frequency can reach an average of 100 frames, but the minimum frame rate (the last indoor scene) is not good. It is not surprising that the average frame number after overclocking has increased slightly, but the minimum frame number has increased by more than 15 in one fell swoop - it can be seen that it is not that the graphics card is full, which means that it is already a bottleneck. Compared with the two comparison groups, the average frame rate is slightly lost but the difference is not big, and the lowest frame rate is "one thing is reduced to another thing".
013 is the year when the next generation architecture Haswell was born. I chose Tomb Raider 9, the most praised new trilogy, with the highest picture quality. The benchmark test of this game is actually different from the actual gameplay, and the pressure on the CPU is relatively low. But only depends on the results. After overclocking the i7-2600K, you can defeat the two comparison groups in one fell swoop, and you can win with the lowest frame rate with a silent frequency. I have to admire the gaming strength of Intel CPU.
In 2014, I chose Middle Earth World: Shadow Mordor , one of the best games of the year, with extremely high picture quality.
This game is very optimized for multi-core. The bulldozer CPU can also drive GTX 1060 level graphics card, so the four players are equally matched. It is the worst one that is R7-5700G... The other two are all above 94, which is 93.7 rounded to 94. Is there too many cores that have fallen into the dispatch pit?
The lowest frame rate is also the winner of Intel. I actually tried to reduce the picture quality by one level, and found that all beings are still equal; the reduction will be a bit out of reality.
015 is another classic architecture. Skylake was born in the year when it was selected for Tomb Raider: Rise , very high picture quality + DX12, anti-aliasing FXAA with a more efficient graphics card. When testing this game, students need to pay attention that the CPU usage of the first benchmark is likely to be too high, so it is best to take the value of the second pass.
From this point on, mainstream games have basically bid farewell to the optimization of "ancestral flyer core". Multi-core strong CPUs are easy to run higher, so I am still restricted by graphics cards. After overclocking the i7-2600K, it is basically the same as the two comparison groups. It is necessary to count to 0.1% low before it is significantly behind; if it does not overclock, it will suffer a little, with an average of 6 frames lower.
In 2016, I chose Far Cry: Original Killing , very high picture quality.
This game is a new turning point in the configuration requirements (in the previous Great Revolution and the Wizard 3 were actually the same), and GTX 1060-level graphics cards cannot be turned on at the highest quality. After reducing the image quality for one level, the four players still couldn't break the gap. Only the silent frequency group was limited in single core and bandwidth, and the minimum frame rate was lost a little more.
017 is the birth of Zen (after that Intel bid farewell to the "toothpaste squeeze" journey), and chose Assassin's Creed: Origin , high picture quality.
The game was just released in Zen. The 7th-generation Core still maintains quad-core was a rare CPU killer back then. The next generation of Odyssey continued to work hard and won the reputation of "Equality of All Beings", but in fact its origin is almost the same as that of Odyssey. It has become a rare gap in the "three-level jump" in this test - the
- i7-2600K silent frequency can only run to an average of 45 frames, and the lowest will be 30; the improvement after overclocking of
018 selected Tomb Raider: Shadow , high picture quality + DX12, anti-aliasing selection TAA.
This game is actually quite CPU-intensive. After overclocking, the i7-2600K is increased by 6 frames compared to the default frequency, which is quite obvious in terms of amplitude. Interestingly, the minimum frame rate of single core difference in this year should be a disadvantage, but the minimum frame rate of the two protagonist groups is still better than the two comparison groups, which is very strong.
In 2019, the selected Total War: Three Kingdoms , medium-quality.
This game belongs to the strategy type, contains a large number of AI operations and may use the AVX2 instruction set. In the end, the two protagonists tied with the bottleneck of the graphics card, and the 7700K was overclocked with the small negative fake; the R7-5700G full body was the only one with its strongest CPU performance, and it was more than 10 frames higher than the protagonists.
020 is the tenth year when i7-2600K was released. The choice is Cyberpunk 2077, medium-quality image.
The game now sets the FSR gear to "auto" in medium-quality quality. Fortunately, I actually verified that the four contestants should all run according to super quality (that is, the image quality is kept the most complete). The average frame rate is only i7-2600K, and the minimum is equal to all living beings, and the bottleneck of the graphics card is very obvious.
The last four are all popular games: LOL, GTA5, chicken eating, eternal disaster, among which eternal disaster is already the game in 2021. Their common feature is that there is no benchmark or the related functions are incomplete, so they manually record their respective frame performance using MSI Afterburner.
League of Legends has the highest image quality. Students who have seen more reviews on my laptop should still remember that it is a three-slow, so the 8M L3 i7-2600K will definitely not beat the 16M R7-5700G.
However, the performance/memory bandwidth/latency of the three-snap is still effective. After overclocking, the average frame rate of i7-2600K is increased by 50 compared to the default frequency; the minimum frame rate exceeds the average frame rate of the default frequency.
Grand Theft Auto 5 has very high image quality. The i7-2600K silent frequency cannot drive the RX 590 to fully play, and the latter is snatching the fish most of the time. After overclocking, the average frame rate caught up with the two comparison groups in one fell swoop, but the minimum frame rate was still lost a lot. The graphics card was still idled during large-scale car accidents. Considering that this game is from the PS3 generation, the mainstream i5 was only 4 cores and 4 threads when it was released. I think its optimization is not as good as the legend.
PUBG The picture quality is three extremes. At this time, the R7-5700G can't make the RX 590 always fully loaded, and the graphics card has no bottleneck. The performance of the
i7-2600K silent frequency is naturally not very good. The average frame rate is 120+ and it looks OK, but the minimum frame rate can be as low as 72, which is quite affecting emergencies. Because the absolute value of the frame is high and the variable control I played by hand is not good, the overclocking group + the two control groups can basically be regarded as a considerable performance - this game is not the legendary "eating memory bandwidth", but the delay is more common. DDR3 2133 catches up with DDR4 3600 with lower latency.
Eternal Journey is very unfriendly to A cards. RX 590 can only run at low (not the lowest) picture quality, and the medium picture quality has dropped to 40~50.
Low image quality + TAA anti-aliasing setting, i7-2600K also suffers from a silent frequency, and the minimum frame rate is unstable. After overclocking, the average frame number and the minimum frame number have caught up with the two comparison groups, and they have reached the level of mutual victory or defeat. If you are equipped with an RTX 3050-level graphics card, you should be stress-free when playing with 1080P high-quality image quality.
Productivity
The previous game tests are particularly surprising - without deliberately showing up, i7-2600K is not inferior to the juniors who have been released ten years late, with a doubled scale, and more than one-third of the single core strength, most of the time! So what happens to work? Let’s first use professional software with high intensity.
emmmm, it seems a bit not very good...
PS As we all know, it eats memory capacity, but it also has 16GB of memory. i7-2600K loses a lot compared to R7-5700G. Even though the CPU memory double-exceeding results are still not ideal, it has not reached the level of i5-1135G7 thin and light notebook + core display many years later - This should be the biggest pain in DDR3 memory . There is a bottleneck to use
A card to use PR, but it still fails miserably while the opponent also uses the same graphics card.
AE simply ran to halfway and reported an error and aborted. It is guessed that the Adobe CC version I used in the evaluation is too new. It may require some features that are not supported by the second generation Core, and it must be replaced with the old version. The situation of
SPECviewperf is much better, and you can return to the state of winning and losing between each other, and you can even play a single hit even if you have a silent frequency. After all, I have mentioned in the previous notebook review that the configuration requirements are approximately equal to 3A games.
Look at the low-intensity PCMark10 Modern office ... It seems that it is still not very good. The overall improvement after overclocking of
i7-2600K is very large, with a total score of 4809 of the silent frequency, which is 20%. In actual operation, it is obvious that the speed of the switching software is faster than that of the silent frequency, but compared with R7-5700G, it is still helpless to lose.
Of course, it can be seen that more projects here test single cores. The false 7700K overclocking performance, which cut half of the cores and slightly reduced frequency, is still strong.
Power consumption/temperature
Here you use the default frequency to display power consumption (I forgot to change the memory overclocking back).Due to different heat dissipation conditions, the temperature of the comparison group is not put in. Although
i7-2600K is an old 32nm process, thanks to Intel's excellent turbo frequency training, its power consumption is not high under light load, and the desktop is only about 10W during standby time; even if I press the top to overclock it to 4.8GHz of the full core, it is only increased to 35W. It should be able to lower it with Offset mode.
opponent R7-5700G is of course more advantageous because it is derived from a laptop. But if it is a traditional desktop CPU with independent IO dies like R7-5700X, it will be much higher. Even TSMC 7nm (IO die is 12nm in GPU) that is at least 3 generations new to Intel 32nm, it can be close to 30W.
When playing 1080P video locally (both selected for hard resolution of graphics card, the same below), the power consumption of i7-2600K slightly increased to 19W; after overclocking, it is 37W.
Bilibili watch 1080P video online. The power consumption after opening the barrage is 33W, which is still limited to the opponent; after overclocking, it is 48W. In the
FF15 benchmark test, the scene with the highest power consumption of i7-2600K is actually a chorus rather than a battle. However, in order to unify the opponent R7-5700G, it only takes the maximum value during the battle (so screenshot cannot be taken). The maximum silent frequency is 72W, but it is not particularly exaggerated after overclocking. The settlement scenario where
benchmark runs, the power consumption of 47W of the default frequency, and overclocking is slightly higher than 56W. The opponent's 5700G is very good as an APU game power consumption control, but traditional desktops are very tragic. For example, if the 5950X is turned on to play games, its power consumption can even exceed the label TDP. When the i7-2600K package consumes 55W, even lower than the opponent (called you have more cores) when the CPU is single-baked (non-FPU, only using integer units). When overclocking, it only increases to about 70W.
single baking FPU, which has been shown before overclocking. The i7-2600K operates stably at 3.6GHz for the entire core, with a working voltage of 1.311V, and the packaging power consumption of 92W is not as good as the TDP label value; the room temperature reached 27℃ when I tested, but under the pressure of Yajun G3, it was still only 62℃, which was very cool.
Conclusion
has been around for more than 10 years, and i7-2600K is still beyond imagination today! Especially under the dual CPU and memory, facing the R7-5700G, which is ten years late, doubled in scale and more than one-third of the single-core strength, the gaming field is still not inferior - you may say that the graphics card I chose is low, but how many people will use it that only supports PCIe2.0 with 3060Ti and 3070? In fact, it is enough to have a 3060/6650XT at 2K resolution.
Of course, the premise is that you only play games. When we come to the work field, the i7-2600K is still not very strong single-core performance after overclocking, and the small scale of 4 cores and 8 threads, and lacking the support of DDR4/AVX2, it is still very difficult. In fact, even for daily use + desktop office, the actual test of PCMark10 is much worse than that of current mainstream CPUs, but it is a bit beyond the range of human perception. You can experience it by changing 13 seconds to 10 seconds, but it may not be possible to change 1.3 seconds to 1 second.
But don’t forget, i7-2600K is 11 and is about to be 12 years old. When it was released, the iPhone was only released to 4, and Weibo and WeChat were just popular. Yao Ming is still an active athlete, and Jay Chou’s voice is still at its peak. From the perspective of gaming consoles, a PS3 generation CPU is still in battle when it comes to PS5, which is very unimaginable - Perhaps Intel really did not "squeeze toothpaste". It was because the starting point of Sandy Bridges back then was too high, which led to the difficulty of improving the subsequent generations of CPUs .
i7-2600K As a legendary god U, it still survives today.
stared at the 13th generation Core, who was ready to go. Thinking back to his heroic posture back then, it must have a smile on its mouth.
This article ends here! There will be the next article for the second question at the beginning.
I am Yuna, a hardware and digital enthusiast. Looking forward to your next reunion~~
