Hi Frankz,
Ok, now I understand, thank you. Yes, in studying those CPUs, I can see clearly that three of them would offer you a substantial upgrade:
| Host CPU | Cores / threads | Turbo | TDP | PassMark CPU Mark | Single-thread bench |
| i7-9700 | 8 /8 | 4.7 GHz | 65 W | 13,175 | 2,744 |
| E-2224G | 4 / 4 | 4.7 GHz | 71 W | 7,526 | 2,688 |
| E3-1225 V6 | 4 / 4 | 3.7 GHz | 73 W | 5,896 | 2,151 |
| i5-4670 | 4 / 4 | 3.8 GHz | 84 W | 5,484 | 2,125 |
Passmark CPU Benchmark shows your aggregate performance benchmark would jump from around 18,000 points to 39,000 points. Not to mention the most important part which is that you'd be doubling your physical core count. What's really neat is that you actually will likely end up saving electricity too:
- Old cluster:
73 + 84 + 71 = 228 W
- i7 cluster:
65 × 3 = 195 W
But that's the TDP, not idle power consumption.
The last benefit is your homogeneity will permit VMs to live migrate with a lower risk of crashing... But on the other hand your biggest penalty is losing ECC memory which overall increases the risk of crashing even at idle. Every day, gamma rays rain from the sky and flip bits on electronics and storage media. ECC is one of the most important tools to overcome this. But honestly for a lab it's overkill anyway.
Enjoy, sounds like a good lab setup.
Tmanok