At STH, we have been big fans of the Intel Core i3-N305 for some time. It was the first SoC in its class to offer eight cores, providing a big performance jump over previous generations. That was a 2023 part, and in 2025, we got the update in the Intel Core 3 N355. This was an upgrade, but one that is worth at least taking a moment to look at.
The Intel Core 3 N355 Update Over the Core i3-N305
The Intel Core i3-N305 was released in Q1 2023. Here is a look at the lscpu output from the 2023 era 4x 2.5GbE and 2x 10GbE Intel Core Firewall and Virtualization Appliance.
Last week, we looked at the new Qotom 10Gbase-T Mini PC with the Intel Core 3 N355. Here is that lscpu output.
In either case, you can see the eight core CPU complex. For those who are a bit more visual and want to see the cache layout with the cores, here is the topology from the iKoolCore R2 N305 machine:
Here is the N355 version. You can see that we have the same 6MB of L3 cache and then two four core clusters each with 2MB of L2 cache.
That Intel Core 3 N355 is a Q1 2025 processor, or two years after the N305. If they look similar, that is because they are. Here is the Intel Ark comparison.
Both CPUs have the same:
- Core count
- Cache capacity
- TDP
- Single channel memory
- Memory speed
Perhaps the biggest change on the CPU side of the SoC is the clock speed. The N305 is a 3.8GHz maximum turbo clock speed part while the N355 is a 3.9GHz maximum part. While that is around a 2.6% clock speed increase, the big question we had was whether that translated into a 2.6% performance increase.
This is just a “look, we’re not totally inactive” part
How can this possibly be strange to a CPU reviewer? It’s the same silicon. Are you missed by the marketing name?
If it were the N306, you wouldn’t write articles bemoaning 1-3% change when the only difference is a 100mhz turbo bump.
I (used to) expect better out of servethehome. Patrick, does this really help your long term value as a site?
Gutless to complain anonymously.
This article very clearly says that both are the same silicon and just have a different clockspeed. Based on that nobody is surprised.
Still it’s fair that John complains about the only 1-3%. It is expected due to what Intel did, but that two years later we just get a refresh instead of a real new generation is disappointing.
“100MHz at 3GHz would be a 3.3GHz gain”
wut
I think
-100MHz at 3GHz would be a 3.3GHz gain
+100MHz at 3GHz would be a 3.3% gain
Yes, please revise first paragraph of page 2.
The lstopo image shows the PCIe lanes go to the CPU, but if I remember correctly, those PCIe lanes are connected to an integrated chipset which connects to the CPU through an internal bus called OPI (on package DMI). Has anyone verified what is the bandwidth available through OPI?
Time for some thermalpaste hygiene ! :-)
cw824 and michaelp – Yes. Swapped to percentage. I am behind quite a bit today since I was up at 6:30am to get the HPE Keynote at the Sphere and am sitting down for my first time
Not given – This is the 2-year refresh. Intel has done these before, but I thought it made sense that John wanted to point out that in two years, this segment is only seeing a 1-3% increase. To be fair, we are at 9 quarters at this point, so it is more than two years. Some may see a N305 and N355 and think they are getting something much faster, when in reality, it is such a small difference that it is hard to perceive or even measure in most cases. Remember, there are folks who do not keep up with the industry every day and the N300 to N305 difference is greater than a 10% performance delta. If N300 to N305 was 10%, one might think that N305 N355 is more than 1-3% just based on that model number.
I think that is a very fair point.
cw824 and michaelp – Yes. Swapped to percentage. I am behind quite a bit today since I was up at 6:30am to get the HPE Keynote at the Sphere and am sitting down for my first time for 5 min today.
Not given – This is the 2-year refresh. Intel has done these before, but I thought it made sense that John wanted to point out that in two years, this segment is only seeing a 1-3% increase. To be fair, we are at 9 quarters at this point, so it is more than two years. Some may see a N305 and N355 and think they are getting something much faster, when in reality, it is such a small difference that it is hard to perceive or even measure in most cases. Remember, there are folks who do not keep up with the industry every day and the N300 to N305 difference is greater than a 10% performance delta. If N300 to N305 was 10%, one might think that N305 N355 is more than 1-3% just based on that model number.
I think that is a very fair point.
To me it seems like Intel did an update to include Intel® Trusted Execution Technology ‡, as it seems like this isn’t in the N305 chip. I don’t know how relevant this feature is for most people but it seems like the only change besides a very minor bump in speed so I assume that at least someone at Intel believe it is an important feature…. I wish the article would address this feature change – seems like it has been completely missed ?
Yah, it does seem like the marketing department is just being overly active, thinking that people will choose the new part and hopefully pay more for it, simply because 355 > 305. But of course, in reality, the difference is going to be imperceptible.
I mean, the bottom line is that for most use cases, either is going to be fine and just buy the cheaper one, which is exactly what the article concludes.
Here we go again, Intel produce low end garbages increase insignificant speed improvement, more like a stealthy CPU dumping into market help bump up it’s hardly sold high-end CPU bait in AI demanding age.
It’s not a refresh, which refers to a change in silicon, often the same microarchitecture ported to a new node, or at least with a few optimizations to take advantage of the existing node.
It’s a rebadge of the same silicon on the same process.
It’s not a refresh, which refers to a change in silicon, often the same microarchitecture ported to a new node, or at least with a few optimizations to take advantage of the existing node.
It’s a rebadge of the same silicon on the same process.
Bystander/ Not Given – Intel uses refresh not for changes in silicon and new nodes, but for SKU updates in an existing line. Here you go https://www.servethehome.com/big-2nd-gen-intel-xeon-scalable-refresh-brings-competition-anew/
And what about thermals of CPUs?