The IBM Processor Value Unit table, updated.

The PVU table is the lookup that turns a processor into a price. Every PVU calculation starts by finding the right row for your chip, and the wrong row quietly scales your entire position. Here is how to read the table correctly and which rows tend to be misapplied under audit.

The Processor Value Unit table is IBM's published mapping from processor family and model to a PVU per core rating. The PVU formula is only as right as the row you pull from it: PVU per core times allocated cores. Choose the wrong rating and the error is not additive, it is multiplied across every core in scope, which is why a single misread row can move a position by hundreds of PVU.

How the table is organized

Ratings are assigned by processor type, vendor, brand, and model number, not by a single universal figure. As a working reference, a common Intel core sits around 70 PVU, while higher throughput processors such as Power carry higher ratings and some lower powered chips carry less. Because the rating depends on the exact model, identifying the processor precisely is the first and most consequential step. A family level guess is where most rating errors begin.

How to read a row correctly

• Identify the exact processor. Capture vendor, brand, and model from the hardware inventory, not a generic family label.
• Match to the precise row. Find the table entry for that specific model rather than assuming a representative number for the family.
• Apply the rating to allocated cores. Multiply the per core rating by the cores the IBM software is entitled to run on, using sub-capacity counts where the conditions are met.
• Use the table version for the period. An audit lookback can span years, so the rating that applied during each period is the one that governs that period.

Where the table is misapplied under audit

The recurring errors are mismatches between the hardware and the row. A processor is assigned a family default instead of its model rating; a refreshed server keeps the old rating after a hardware change; a mixed estate is flattened to one assumed figure; or a high throughput chip is rated as if it were a commodity core. Each of these is checkable against the inventory, and each can swing the position in either direction. Our reconciliation rebuilds the rating for every processor in scope from the hardware record and the table version that applied, so the multiplier at the base of the calculation is defensible before anything is added on top.