Discussion Just how usable is a 9 year old CPU for me? - My experiences using an i7 2600k in 2020 for a week

Page 2 - Seeking answers? Join the Tom's Hardware community: where nearly two million members share solutions and discuss the latest tech.
Backstory:

I am currently occupying myself with flipping computers. I have found that many buyers will pay a lot of money for a computer with an "i7" even if the older i7 is worth as much as a much newer i3. I figured I should force myself to live with one to see if I should sell computers with older i7 paired with expensive graphics cards as "gaming computers" without feeling dishonest.

I recently bought the i7 2600k+cooler+ram+mobo bundle below for $175+$9 shipping. Currently, I am on day 4, though I only used the PC for a couple of minutes of web browsing one day.

Here's how it is going so far. I will update this thread for the next few days as I do more and more with the system.

System specs:
Intel Core i7 2600k
Thermaltake UX100 65w RGB cooler*
Asus P8Z60-V Gen3
16gb (4x4gb) Crucial DDR3 1600mhz
MSI Mech OC RX5700 8gb GDDR6
Kingston A400 240gb SSD
Corsair CX550M bronze
NZXT S340 with a single exhaust fan and side panel off

*I was impressed by this 65w cooler as it kept this 95w CPU at around 60-70c in games. Even under prime 95 for awhile the CPU hovered around 79/80c. This would not hinder the CPUs performance in any way. Turbo was functioning correctly. I would assume the reason this cooler performs well is that the fan uses a 3 pin 120mm fan that runs at full speed 100% of the time. This helps keep the CPU cool but has the side effect of being loud even at idle.

The RX5700 is a mid-range GPU performing superior to the RTX2060. This seems like a lot to pair with a 9-year-old CPU, but we will see.

First impressions and general usage performance:

My first impressions of the i7 2600k were strong. The system bootup time is fast and the CPU is very fast for web browsing. Some of this can be attributed to the SSD and 16gb ram, but the CPU handled itself nicely.

Performance for my favorite games:

I personally game at 1080p high detail. I am happy with a solid 60fps that is stable.

First up, Automation
The performance of this game disappointed me. The game hovered around 35fps and dipped a lot at 1080p maximum with a 150% res scale. lowest settings and 50% res scale did not even touch the framerate as in every case the GPU was waiting around on he old i7. The GPU never maxed out and oddly enough the CPU did not seem to have any cores hitting 100%, which is weird. Also, when I clicked on any buttons in the game it would cause lag for one second, which seems common across the low-end CPUs I have tested this game on. This game is not competitive so setting a 30fps lock made the experience playable, not great.

Next up, BeamNG.Drive
This game performed better, quite admirably. This game's performance is very very dependent on the map and the number of cars on screen. It is very physics heavy on CPUs.

Even on a very demanding map of Italy, I was able to manage around 60fps on average, though in heavy city areas the framerate would dip to the 50s, and in more rural areas the framerate would go into the 70s. I do not usually have more than one car on screen at a time, however for other people that load in several cars, the performance may suffer. Every other map managed even better performance, for example, Hirochi Raceway managed between 80-90fps or even more. Grid map would likely get hundreds of FPS.

Overall, very satisfactory performance, however, the RX5700 mainly sat around waiting for the CPU.

GTAV
GTAV online offered very similar performance to BeamNG.Drive. At 1080p max (except no msaa) the i7 averaged around 60fps overall. In the city, I got roughly 50-60fps, usually closer to 60 than 50. In the desert, I got over 60fps most times, though sometimes it would dip below 60s for a little bit. The action would dip this framerate. Changing the settings to 1080p bare minimum did not touch the framerate as in all cases at 1080p the RX5700 was waiting on the 2600k. I noticed several cores bouncing off of 100% however it was not all of the time. Setting this game to a 60fps lock seems like a good idea.

Banished
Banished is not demanding at all. I have run this on a Celeron netbook + 2g ram with a rather not good experience, but it actually ran. On the i7 2600k with RX5700 I actually saw the first instance of a GPU bottleneck, as much as I hate that term. At 1080p max I got several hundreds of FPS, however, the exact framerate varied a TON depending on how zoomed in the game was and how fast it was sped up. Reducing settings to lowest gave me a generally higher framerate, showing the framerate was not CPU limited most likely. It was odd since no parts seemed stressed at all in MSI afterburner, which seems to happen often on this game with any decent hardware.

So far:
The performance until now has been alright and this old i7 seems to still perform well for games, however, it does not belong paired with a high-end GPU at 1080p. I would bet at 1440p max in many games the 2600k would be an alright match for a GPU like an RX5700 or 2060, however, some games like Automation do not like this i7 regardless.

Upcoming:
I would like to test the performance of this CPU in some other tasks I do occasionally, however, I would also like to try overclocking. This board has an incredible 12+4 phase VRM (likely 6+2 doubled) and large heat syncs, even one on the rear of the board. I will need to find my intel brackets for my Arctic Freezer 34 esports (also has an Arctic F12 PWM added) in order to overclock this CPU well.

I would also like to give my thoughts of the CPU in general VS a newer low-end CPU. I have a feeling a 3300x will end up being recommended instead though it is a little more expensive.


UPDATE - Drawing Some Conclusions:
It seems for 60hz gaming, the Intel Core i7 2600k still performs well if paired with a suitable graphics card and 16gb ram. Esports titles would be a breeze, though more demanding games still can run pretty well, though very demanding titles may dip below 60fps but should still be at least be somewhat playable. Overclocking the i7 2600k to 4.6ghz helped to bring overall framerates up and help to reduce dips below 60fps.

If you currently have an i7 2600k and have a 60hz monitor, you probably do not need a CPU upgrade just yet. If you play at resolutions above 1080p and have a 60hz monitor, a high-end GPU and a decent CPU cooler to overclock might be a worthwhile investment.

If you are planning on building a PC, do not buy a used 2600k.

The CPU itself is priced decently and isn't much more than a new Celeron or Athlon, and the i7 will beat those CPUs. However, the motherboards are getting more and more sparse and seem to be going for over $75 for a cheap z68 or almost $50 for a non-OEM H61, which is about as much as a budget B450 and A320 respectively, which offer features like M.2 and a much better upgrade path. Also, a used kit of 16gb DDR3 1600mhz will cost you around $45, which is about the same as a cheap 16gb kit of ddr4 used.

I paid $175+9 shipping for the bundle of
i7 2600k
brand new Thermaltake UX100
Asus Z68
16gb ram
For the same money new you can get
PCPartPicker Part List

Type|Item|Price
:----|:----|:----
CPU | Intel Core i3-9100F 3.6 GHz Quad-Core Processor | $74.99 @ B&H
Motherboard | Asus PRIME H310M-E R2.0 Micro ATX LGA1151 Motherboard | $56.07 @ Amazon
Memory | GeIL EVO SPEAR 16 GB (2 x 8 GB) DDR4-3000 CL16 Memory | $57.99 @ Newegg
| Prices include shipping, taxes, rebates, and discounts |
| Total | $189.05
| Generated by PCPartPicker 2020-06-04 10:07 EDT-0400 |
Sure its slightly more money and you miss out on overclocking, get a bad stock cooler, and the cpu has only 4 threads, but its far more modern. It has a better upgrade path, the motherboard supports m.2, it has an upgrade path to 6 or 8 cores, and honestly on par if not superior CPU performance. If you buy those parts used, it will cost similar to the i7 bundle.
 
Last edited:
Ok, I am having issues with timings.

Regardless of clock speed, I cannot adjust the timings to 12-12-12-36 as that causes no boot. Even leaving all of them at stock but changing just the first one to 12 causes no boot.

There are a bunch of sub timings below it I suspect need to be changed, but I have no clue what to do.
 
Interesting read. The office used PC market is flooded with 2000 series intel processors. You have to make an effort to find more modern processors. That might speak to intels utter lack of self improvement during those generations. Our office uses HPs with i5-2500s. No GPUs. They get work done.
 
So Cinebench scores.

Current 2600k setup:

4.7ghz 1600mhz cl11 ram
1764cb multi-core - 348cb single core
4.6ghz 1866mhz cl11 ram
1746cb multi core - 343cb single core.

Intel Core i7 2600k
4 Heatpipe RGB triple fan Chinese cooler
MSI Z68A-G45 (B3)
16gb (4x4gb) Corsair Vengence DDR3 1600mhz
Samsung 850 EVO 500gb SSD
Zotac Mini GTX1070 8gb
EVGA GD600w 80+ GOLD
Raidmax NEON ARGB ATX mid-tower case
Windows 10 Pro - Activated

Around 4.4ghz with 1600mhz ram
325cb single core - no multi core tested.

  • AMD Ryzen 5 2600 @4.2ghz 1.3625v
  • Arctic Freezer 34 eSports Cooler
  • MSI B350 Gaming Plus (Bios - 7A34vMF)
  • G.skill Aegis 2x8gb Samsung b die overclocked to 3466mhz CL18 @1.45v
  • MSI Mech OC RX5700 8gb GDDR6
  • Kingston A400 240gb SSD + 500GB Samsung 5900RPM
  • Corsair CX550M 80+ Bronze
  • NZXT S340 Red&Black + 2x Arctic F12 & 2x Arctic F14
  • BenQ GW2270 VA 1920x1080 60hz |
  • Windows 10 Pro
At the time of that benchmark, I had an RX480 and hand bought my RX5700 yet. /SPOILER]
for this benchmark I cranked up the clocks to 4.3ghz and ram to around 3466 cl18 for that benchmark. Normally at 4.2ghz 3200mhz ram
I got 445cb single core in r20. I cannot find my screenshot of multi-core.
 
Might be margin of error or its unstable timings, i would test memory benchmarks to see how much you gain from 1600 to 1866mhz, you might get better results staying on 1600mhz and getting as low timings as you possible can, aim near 9-9-9-24

Free trial of aida64 has memory benchmark.

EDIT: seems like gains from 4.6 to 4.7 are very small and you can use much lower voltage to keep it stable. Its also better for the internal memory controller which will be cooler since very high vcore can affect imc temperature because its located in the cpu chip. Then you can use slightly higher VTT and still get decent temps on the IMC (which cant be monitored no thermal sensor for imc)
However you usually dont need to raise the VCCIO / VTT much if you dont raise ram speed over 1866mhz, 1.15v should be fine. DRAM voltage might help with timings even when frequency is lower.
 
  • Like
Reactions: SamirD
Its also the northbridge frequency which matters alot of overclocking with p55 h55 chipset, maybe its same with your gen. You cant change it in bios so when you increase BLCK you will automatically increase nb freq,

With 200blck im getting 3600mhz on the Northbridge and usually thats the high limit for this generation northbridge. Some boards can get issues even with 3450-3600mhz. Cpu can also be a factor in this.

Getting higher nb freq also increases your performance so getting good balance is the key. When i increase to 210 blck i can still run the system (almost 3800mhz nb freq) but need to use crazy vccio voltage of 1.4v which might kill the imc inside CPU in the long run. Im currently settled for 200 blck with 1.325v vccio which is still way too high but managable. I also need 200 blck to get ram frequency of 2400mhz so thats why im using these settings.

Can you change northbridge frequency in bios?

edit: im also using passmark memory test and userbenchmark for ram testing. i dont really trust these for cpu benchmark cause they take -30% performance off from older cpu:s just for the lols.
Both sites are also biased, in theyre own way.
 
PCH has 2 voltages, i dont think you need to touch any of these. PCH voltage is just main chip voltage usually default is 1.05v which is fine. PCH PLL voltage is default 1.80v and you do not need to raise it, i seen guides ask you to decrease it with gen2-3 cpu but i have no experience with those.

I only need to raise these when im going over 195Blck and that is only cause of the multiplier limitations of my CPU. You dont have to touch BLCK, i guess 101-103 wont do any harm , its up to you.
 
Well, I definitely used this hardware for more than a week. lol

Today I got some Noctua thermal paste in the mail today and this enabled me to get a little more clock speed, but nothing more than 100mhz before throttling. However, I noticed Cinebench was not as demanding as prime95, so I could get away with higher clock speed.

I bumped the sucker to 5ghz with 1.56v and somehow it ran Cinebench R20 without throttling nor crashing.
The chip doesn't seem to stop clocking higher as long as I keep feeding it more and more voltage, so my main limit is cooling. I also had ram at default 1600mhz CL11. I got a whopping 1902CB, which is far more than what I expected this CPU to pull off. Trying CL11 1833mhz ram didn't touch score at all.

I could not bump the clock speed up anymore since that made the CPU throttle hard and reduce performance.

However, I could try something like 5.1 or 5.2ghz on a single core and see what CB 1t looks like. I would also like to test the CPU at stock to compare Cinebench scores to the 5ghz status.

I have 2 days till the case comes and I put the weak cooler back on it, then sell the sucker, so I will continue to mess with it in these 2 days.

If I had some faster say 2133mhz DDR3 ram as well as a bigger cooler like an NHD15, I probably could top 2000cb. It would be a fun project for sure.