What’s the best processor for a trading computer?
A common question with lots of different answers depending on who you listen to.
My answer is… it depends (useful I know!).
What type of trader are you? Do you sit and watch one chart for hours on end?, or do you run 8 screens with multiple trading platforms, have lots of charts constantly being monitored, all whilst you play Candy Crush and catch up with the latest Netflix series?
What you need is what works best for you as a trader, there is no one size fits all approach.
That doesn’t mean that the process of deciding what you need has to be a nightmare though, far from it, you should be able to look at your requirements, compare different options and then make your right choice.
Based on that idea, we have put together this article which is our CPU / processor mega test.
We have bench-marked the latest Intel processors in a wide range of tests which show not only their raw performance levels but also how that will actually impact your trading software.
As new processors are released we will keep this mega test updated so you can be sure that it will always provide the latest and greatest advice on which is the best CPU for trading.
Test System and Methodology
Our testing computer hardware for all the tests below is exactly the same, the only changes are processors and motherboards.
Different CPU series require different motherboards, for the 7th generation Kaby Lake chips we use the Asus Z270-P motherboards. The newer 8th generation Coffee Lake tests use the Asus Z370-P motherboard, and the Intel Skylake X tests use the MSI X299 Raider motherboard.
All other hardware remains identical across the tests, this includes: 16GB 2,666Mhz Crucial RAM, 240GB Kingston Solid State Hard Drive, nVidia GT1030 2GB Graphics Card, BeQuiet 600w Power Supply, Antec 302 Case, Arctic Freezer CPU Cooler.
In terms of software we used the following: Windows 10 Home Edition, Google Chrome v59, Handbrake 1.07 (64-Bit).
Each test was repeated 3 times and the average of the 3 results was used as the final reported score. This irons out any freak results that may occur and gives a clear insight into average performance levels.
Chart Colour Key
To make identifying results a bit easier I have colour coded the charts below as follows:
- Blue = Intel 7th Generation Kaby Lake Chips
- Green = Intel 8th Generation Coffee Lake Chips
- Orange = Intel Skylake X ‘Enthusiast’ Processors
Web Browser Tests
The first set of tests are web browser tests, these are what we use as proxies for trading software.
The browser tests run a series of simulations replicating processes that many web based trading software applications use as part of their background code, this makes them a great proxy for benchmarking how a particular computer will run a web based trading platform.
Other, none web based trading platforms, like MetaTrader 4 for example, also take data feeds in and perform manipulations on them to create the charts and numbers that you see, so again, these would mirror the type of functions that these browser suites use to benchmark performance.
Our first test is the JetStream test which runs a number of simulations and calculations typical to a modern web application, all your web based trading platforms will use very similar workloads to this.
Interestingly we can immediately see that the Skylake X chips don’t perform very well here, these are touted by Intel as ‘Enthusiast’ processors and there are tons of articles online recommending them (or previous generations of them) as perfect for a trading computer, well this test seems to disprove that straight away.
The two Kaby Lake and Coffee Lake i7 processors top the charts here, closely followed by the Coffee Lake i5 8600K.
While the i3’s are towards the bottom of the listings, they still both beat out the 8 core 7820X despite costing 5 times less…
Google’s Octane browser test is similar to the JetStream one in that it tests workloads similar to what many web applications use, great for comparing to web trading platforms.
A similar pattern occurs here backing up the JetStream test, the Skylake X (high processor core) chips struggle to get near the performance of the i7 8700K, i7 7700K and the i5 8600k.
It’s also interesting to see the new i3 8350K match the performance of the 7th generation i5 7600K despite costing less.
The final test in our web browser series uses the WebXPRT suite, this is a bit different in that it places more of an emphasis on the visual side of things, it renders some 3D images, produces charts and graphs and lots of data tables.
The first thing we see is that the Skylake X chips perform better here but still don’t reach the performance levels shown on the 8700K or any of the other i7 / i5’s. Going on these browser tests alone, it seems hard to justify the cost of these highly expensive processors.
The i7 8700K wins again, as it has done in every web browser test, indicating that it is the fastest processor available for web based trading software platforms.
On this test there is little between the i5 7600K, the newer i5 8600K and the i7 7700K. Looking at the previous web tests shows that out of these three, it seems that the older i7 7700K is the better option.
Let’s now move into some CPU tests to dig down into the raw performance levels of each chip.
CPU Intensive Tests
Our PassMark testing suite has a number of processor tests which aim to clearly identify the strengths of a particular CPU, looking at these tests will give us more of an insight into the specific differences between each of the test processors.
Our first test is the overall score given for each CPU, it is determined by a number of sub tests for processor intensive workloads running directly on the PC. These sub tests include things like number sorting algorithms, calculating prime numbers and encryption workloads.
In a bit of a reverse from the web tests, the Skylake X chips come out on top here, the i9 7900X in particular has a big lead, improving on the 8700K result by 24% which is pretty significant.
On the other end of the scale the i7 7100 really lags behind the rest, the newer 8100 (a direct replacement from the Kaby Lake 7th generation to the Coffee Lake 8th generation) offers an almost 40% performance jump, that’s a massive leap.
Overall the new Coffee Lake chips offer an average jump of over 36% over their direct Kaby Lake predecessors, let’s see if we can find out exactly what’s causing this jump.
This is one of the sub-tests of the previous CPU benchmark, it looks at how fast each CPU is at processing one set of instructions, ignoring multi-tasking performance.
The results here give you an idea of the raw speed of each processor when single-tasking, it’s a good metric to gauge relative performance.
We get less of a spread of results here, again the 8700K comes out on top but the is only a 30% improvement over the worst result of the 8100. That’s a much narrower margin than the 222% of the overall CPU test…
Something else must be propping up those big differences in the first CPU test above.
To measure the multi-tasking performance of a CPU we use a program called Handbrake. Technically it’s a program which converts one video format to another, something which is massively processor intensive.
Handbrake will scale up and use as many processing cores that are available, extra processing cores help spread workloads on a CPU.
So this chart shows the number of video frames Handbrake could process per second when performing the same conversion job on each processor, the higher the number the better.
We can quickly see that this area of performance is where there is real differentiation between the chips.
The i9 7900X destroys the opposition giving a 25% performance jump over the next nearest competitor, the 10 cores on this chip make a massive difference here.
The dual core i3 7100 lags massively behind the field here, all the rest of the chips have at least 4 cores and it shows.
The 8700K is a 6 core chip, 2 more cores than the i7 7700K which it replaces, giving it an almost 44% performance jump.
So, what can we learn overall from these CPU tests?
In terms of multi-tasking performance levels, in particular for software that can scale to use lots of processor cores (multi-threaded software), the Skylake X chips rule the roost.
For raw performance for software that isn’t multi-threaded (most web apps and the majority of desktop based software) then the 8700K is the best option.
There is less of a difference here though so the previous generation i7 and the two i5’s are also strong contenders to consider as well.
2D and 3D Graphics Tests
Moving on to your system graphics, how do these processor changes affect the performance of your graphics engine?
Remember that the graphics card used is identical throughout all these tests.
For trading purposes, the 2D performance of a computer is the most important aspect of the graphics setup, all your charts and trading interfaces are powered by the 2D graphics engine on your graphics card.
You’d assume that having the same graphics card in all the systems would result in the 2D performance being the same across all chips, that’s not the case though.
It seems like the processor does have some impact on the results of this test, although the impact is not too pronounced.
Again, the Skylake X chips perform the worst, this could be down to the different system architecture of their motherboards perhaps.
The i7 7700K tops the sheets here followed by the 8700K and then the i5’s.
For some unknown reason the 8100 system did not like this test, again it’s hard to say why this result was so bad. Digging into it all of the 3 repeated tests were in the same range so it wasn’t just a freak result.
Overall, apart from this 8100 result, the rest follow roughly a similar pattern of improving with CPU performance levels for the Kaby Lake and Coffee Lake processors.
The 3D graphics test is an important test for gamers, or anyone creating / working with 3D models such as CAD engineers.
In terms of trading software, it is much quicker to draw a chart with lines and vector shapes than rendering them in 3 dimensions.
Due to this all trading software uses these simple 2D shapes to represent data to you, meaning that on the whole, the impact of 3D performance is non-existent on how fast your trading software will actually run.
That being said, it’s still worth our time to see if these processors impact 3D graphical performance as you may have other software which makes use of it.
Looking at the results we can instantly see that there is no impact on these when switching from one CPU to another.
The spread between the best and worst result is less than 1%.
RAM & Hard Drive Tests
In your trading computer there are two types of storage, short term storage is where your open programs, files and charts live which is called RAM.
The second type is your hard drive and this is where your files and programs are stored over the long term, this persists when your computer is powered off, RAM is wiped clean on a system restart.
The speed at which your computer can read and write items into RAM will have some impact on your systems responsiveness, does a change in processor affect this?
To be 100% clear, all RAM is very fast now a days, it is rarely a system bottleneck, so this metric is not as important as some.
That being said, we can see that the speed at which your RAM is accessed does tend to improve as we step up the processor series in the Kaby and Coffee lake chips.
Again, the Skylake X systems perform badly here, at a guess I’d say maybe the motherboards for these chips have more of an impact than the processors themselves.
In terms of impact on an overall systems performance the hard drive is sometimes mis-understood.
Moving from an older ‘traditional style’ drive with a spinning disk to a newer solid state drive can have a massive impact on how responsive a computer feels, the quicker the drive can read and write data the faster things like booting up a PC and opening programs happens.
Once a program or file is open and has been transferred into RAM, then the speed of the drive has no impact on how fast your trading computer will run.
So a faster drive can make your system feel more responsive at times but it’s not going to impact how fast your trading software runs once it’s open and in use.
Also, I should say, that after numerous benchmark tests, this hard drive test is the most unpredictable of them all. We can repeat a test 3 times on the exact same system and get three (sometimes wildly) different results.
There is no real progression from one CPU to the other, the overall spread of results is around 13% from worst to best test, but the majority are covered by just a 5% difference, nothing that’s not accountable from standard test variations.
Performance Test Overall System Score
Our Performance Test suite of tools gives out a final system score based on all of the tests combined, this is made up from the CPU, RAM, Disk, 2D & 3D graphics performance tests.
As you’d probably expect by now, the results for the standard series of improvements for the Kaby Lake and Coffee Lake chips.
The i7 8700K is clearly the best option according to this followed by the older i7 7700K and the newer i5 8600k.
The Skylake X processors come out well but where would they be without those big multi-tasking CPU test results?
So, you’re looking to spec out a new trading computer and want to work out which is the best processor for trading?
Based on your trading habits I would propose the following as good places to start:
Scenario 1 – Cheapest Possible Trading Computer Running A Simple Web Based Charting Platform:
I’d say that a processor like the i3 7100 is suitable for this job. The Web Browser tests which simulate trading web applications show that it offers good results that are comparable or better even than the newer i3 8100.
If you were looking to run a lot of charts simultaneously and / or wanted to do a bit more on the computer then I’d say the i5 7600K is a good option.
It will give you a good performance boost with your trading platforms and gives the system greater multi-tasking performance.
Scenario 2 – Good Trading Computer For Use With Multiple Trading Platforms
I think there are two routes to go here, you could stick with the Kaby Lake chips which tend to be better value as they are a year behind the latest 8th generation chips.
If you could get an i7 7700K in budget then this is still a really strong chip which could easily handle multiple trading platforms and a lot more.
Alternatively you could look at the newer Coffee Lake series, here the i5 8600K is a really strong processor which only comes in around 3% behind the i7 7700K on the web browser benchmarks, but improves on multi-tasking performance levels by almost 9%.
Scenario 3 – The Fastest Trading Computer Possible For Use With Multiple Standard Trading Platforms
There is only 1 real choice here, the Coffee Lake i7 8700K, it comes out in front on all the web browser tests, the CPU raw speed test (single threaded) and sits towards the front of the results in all other tests.
For the vast majority of trading software the 8700K will run it faster than anything else you can buy.
Scenario 4 – The Fastest Trading Computer Possible For Use With Multi-Threaded Trading Software
Some back-testing trading software will utilise the extra CPU cores available on a system to improve performance levels, we will put together a guide to these shortly.
Like the Handbrake test above, if (and it’s a big IF) you know you have some multi-threaded software that you want the best performance on then go for the Skylake X chips.
They are experts at powering through these kind of workloads.
The cost of these CPU’s (and their supporting systems) is high though, if you only used multi-threaded software on a fairly occasional basis, or could tolerate waiting a touch longer for your back-testing to complete then again I’d recommend the i7 8700K.
It still offers amazing multi-tasking performance combined with best in class performance on all your standard trading software packages.
That brings to a close our CPU mega test, hopefully you can see the strengths and weaknesses of the various chips on test and how they might affect your new trading computers performance?
If you have any further questions or comments just let me know.
Thanks for reading!